• Categories
• Minería
• Resistencia de materiales
• Geología
• Ciencia
• Naturaleza

Citation preview

Block Caving Geomechanics

E.T. Brown

JKMRC Monograph Series in Mining and Mineral Processing 3

JULIUS KRUTISCHNITI MINERAL RESEARCH CENTRE

THE UNIVERSITY OF QUEENSLAND

Firmado digitalmente por Chichofaim Nombre de reconocimiento (DN): cn=Chichofaim, o, ou, [email protected], c=PE Motivo: Certifico la precisión e integridad de este documento Ubicación: Hyo-Peru Fecha: 2010.04.06 06:54:29 -05'00'

Published by: Julius Kruttschnitt Mineral Research Centre Isles Road, Indooroopilly, Queensland 4068, Australia

Copyright © 2002

Julius Kruttschnitt Mineral Research Centre, The University of Queensland

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic tape, mechanical, photocopying, recording or otherwise, without permission in writing from the publishers.

National Library of Australia Cataloguing-in-Publication Entry: Block caving geomechanics ISBN 1-74112-000-4 L Caving mining. 2. Stoping (Mining). 3. Ground control (Mining). I. Brown, E.T. n. Julius Kruttschnitt Mineral Research Centre. (Series: JKMRC monograph series in mining and mineral processing, No. 3).

Printed in Australia by University of Queensland Print On Demand Centre Cover production by University of Queensland, Brisbane

This book can be ordered directly from the Publisher: e-mail: [email protected] Phone: Fax:

+61 73365 5888 +61 733655999

ii

FOREWORD

It is important that research outcomes be disseminated in a useful form to the clients of the

research, and to the community at large if appropriate. The research monograph is a traditional mechanism for reporting substantial bodies of research which, taken together, advance the field to a significant degree. In 1996 the JKMRC published two such monographs, on comminution and blasting, in a series on mining and mineral processing. The present volume continues the series. Caving is a mining method which is of growing interest to companies concerned with the exploitation of massive ore deposits, because of its low cost. However the body of theoretical and practical knowledge of the factors controlling the caving process, particularly in competent rock masses, is limited, and the economic risk in developing a caving mine can therefore be higher than one would like. To remedy this situation, a number of major mining companies came together in 1997 to fund the International Caving Study, a wide-ranging research and technology transfer project. This book records in part the results of the first phase ofthis study. The JKMRC as the lead researcher in the rcs has been fortunate in its collaborators: the nine companies who funded and directed the work, its research partner the Itasca Consulting Group, Dr. Denis Laubscher, and a number of other consultants. An these are acknowledged in detail in the book. On a personal note, I would like to recognise the good work done by the JKMRC research staff and postgraduate students under the able direction of Dr. Gideon Chitombo. Finally, we thank our distinguished colleague Professor Ted Brown for taking on the demanding task of bringing the book to fruition, which he has completed with his customary skill and energy. I hope that the mining and geomechanics communities will find the book of interest and value.

T.J. Napier-Munn Director - JJO.tlRC

iii

ACKNOWLEDGEMENTS

This book is an outcome of the International Caving Study Stage I carried out in the period 1997-2000 by the Julius Kruttschnitt Mineral Research Centre (JKMRC), The University of Queensland, Brisbane, Australia, and the Itasca Consulting Group, Inc, Minneapolis, USA. The following sponsoring companies and their representatives who monitored the progress of the Study are thanked for their support of the Study and the preparation of this book: De Beers Consolidated Mines Limited CODELCO-Chile Newcrest Mining Limited Noranda Inc. Northparkes Mines PT Freeport Indonesia Rio Tinto Limited TVX Gold Inc. This book draws heavily on the original work carried out for the International Caving Study Stage I by researchers from the JKMRC and Itasca, consultants to the Study and several JKMRC research students. I wish to acknowledge, in particular, the important contributions made by: •

Dr Gideon Chitombo, JKMRC, who was the guiding force behind the Study and arranged for me, and this book, to be part of it. He wrote the first drafts of Section 4.3.6 and, with the assistance of Italo Onederra of the JKMRC, wrote Section 5.6. He made a number of valuable suggestions about the contents of several other chapters, provided a great many pieces of information that are included in the book, prepared the initial version of the index and managed the arrangements for the book's production;

•

Dr Bob Trueman, JKMRC, who supervised and personally carried out much of the work reported in Sections 3.1,3.2,3.3 and 5.5, and substantially wrote those sections;

v

•

Dr Loren Lorig, Itasca Consulting Group, who carried out the analyses for, and wrote the original versions of, Sections 3.5 and 3.6, Appendix B and, in conjunction with Dr Peter Cundall of Itasca, Appendix C;

•

Matt Pierce, ltasca Consulting Group, who, with Dr Bob Trueman and Ridho Wattimena, carried out the numerical analyses reported in Section 5.5;

•

Dr Geoff Lyman, JKMRC, who carried out the original analyses for, and wrote the initial version of, Section 2.6.3;

•

JKMRC PhD students Neal Harries, Clare Mawdesley, Brian Eadie and Ridho Wattimena whose research work made significant contributions to Chapters 2, 3, 4 and 5, respectively;

•

Alan Cocker, JKMRC, who developed the software for the JointStats system reported in Section 2.5.7;

•

David La Rosa, JKMRC, who developed the software for the CaveRisk system reported in Section 11.5;

•

Dr Dennis Laubscher whose Block Cave Manual, including the contributions made by Nick Bell and Glen Heslop, provided an invaluable source of information, ideas and illustrations, many of which appear in the book; and

•

John Summers, CGSS, Berkshire, England, who, with input from Dr Gideon Chitombo and others, developed the CaveRisk system described in Chapter 11 and wrote the report on which that chapter is based.

In October, 200 I, copies of the first draft of the book were distributed to the sponsors of the International Caving Study Stages I and 11 for comment. I am grateful to the representatives of the sponsors for their support in this final stage of the process. I would especially like to thank the following individuals for having provided valuable comments on parts of the draft and/or additional material for inclusion in the book: •

Richard Butcher, WMC;

•

Joaquin Cabello, Golder Associates;

•

Dr Gideon Chitombo, JKMRC;

•

German Flores, Chuquicamata Division, CODELCO-Chile;

•

Dr Antonio Karzulovic, A Karzulovic & Associates;

•

Craig Stewart, Northparkes Mines; and

•

Dr Duncan Tyler, Newcrest Mining.

vi

I also wish to thank those who gave their support and assistance to this undertaking in a number of important ways, especially: •

Libby Hill, JKMRC, who undertook the desktop publishing with her usual skill, grace, and efficiency;

•

Vynette Holliday and Naomi Mason, JKMRC, who assisted Libby in this process by preparing many of the figures;

•

the former Manager of the Dorothy Hill Physical Sciences and Engineering Library, University of Queensland, Gulcin Cribb, and Library staff member, Diana Guillemin, for their assistance in providing copies of a large number of sometimes obscure references;

•

Rob Morphet and the partners and staff of the Brisbane office of Golder Associates Pty Ltd for providing me with facilities, encouragement and support during the writing of parts of the book;

•

John Markham, CEO, Itasca Consulting Group, for his efficient project administration; and

•

my partner, Dr Dale Spender AM, for her continuing tolerance of my interest in holes in the ground and for understanding that "the Earth sucks".

Finally, but most importantly, I should like to acknowledge my debt of gratitude to the foundation Director of the JKMRC, Professor Alban Lynch AO FTSE, for inviting me to become involved with the work of the Centre when I joined the staff of the University of Queensland in late 1987. I also wish to record my appreciation to his disciples and successors as Directors of the Centre, Professors Don McKee and Tim Napier-Munn, who have continued to make me welcome at the JKMRC in the intervening years. Without their friendship and support, I would not have had the opportunity, or been able, to prepare this book.

E TBrown Brisbane 29 March 2002

vii

CONTENTS

.FOREWORD ......................................................................... iii ACKNOWLEDGEMENTS ....................................................... v CONTENTS ......................................................................... viii

CHAPTERl INTRODUCTION

1.1

1.2

1.3

lA

1.5

UNDERGROUND MINING METHODS ................................................................... 1 1.1.1 General Features ........................................................................................ 1 1.1.2 Classification of Underground Mining Methods ............................................... l BLOCK AND PANEL CAVING .............................................................................. 3 1.2.1 Outline of the Method ................................................................................ .3 1.2.2 Basic Caving Mechanics .............................................................................. 8 1.2.3 History of Block Caving ............................................................................ 12 BLOCK AND PANEL CAVING OPERATIONS ...................................................... 16 1.3.1 Overview................................................................................................ 16 1.3.2 El Teniente Mine, Chile............................................................................. 16 1.3.3 Premier Diamond Mine, South Africa .......................................................... 20 1.3.4 Henderson Mine, Colorado, USA ................................................................ 24 RISK IN CAVE MINING ...................................................................................... 27 1.4.1 Risk Factors ........................................................................................... . 1.4.2 Introduction to Risk Assessment ................................................................. 29 SCOPE Al'l"D CONTENTS OF THIS BOOK ............................................................. 30

viii

CHAPTER 2 ROCK MASS CHARACTERISATION

2.1 2.2

2.3

2.4

2.5

2.6

2.7

2.8

2.9

DEFINING THE MINING ENVIRONMENT ....................................................... 32 GENERAL DATA REQUIREMENTS ................................................................ 33 2.2.1 Geology ............................................................................................. 33 2.2.2 Surface and Groundwater Hydrology ....................................................... 35 2.2.3 Topography and Environmental Constraints .............................................. 35 2.2.4 Geotechnical Studies ............................................................................ 35 CLASSIFICATION AND DESCRIPTION OF DISCONTINUITIES ........................ 36 2.3.1 Classification ...................................................................................... 36 2.3.2 Description ......................................................................................... 41 DISCONTINUITY DATA COLLECTION BY DRILLING, CORE LOGGING, DOWN-HOLE SURVEYS, SCANLINE AND CELL MAPPING ............................ .42 2.4.1 Introduction ........................................................................................ 42 2.4.2 Geotechnical Core Logging ................................................................... 43 2.4.3 Exposure Mapping Methods .................................................................. 49 ANALYSIS AND PRESENTATION OF DISCONTINUITY DATA ........................ 55 2.5.1 Introduction ........................................................................................ 55 2.5.2 Error and Uncertainty in Discontinuity Analysis ........................................ 56 2.5.3 Discontinuity Orientation Analysis .......................................................... 58 2.5.4 Discontinuity Frequency/Spacing (Intensity) Analysis ................................ 60 2.5.5 Discontinuity Persistence (Size) Analysis ................................................. 64 2.5.6 Definition ofGeotechnical or Structural Domains ...................................... 66 2.5.7 JK Jointstats Discontinuity Data Management System ................................ 67 SIMULATION OF ROCK MASS GEOMETRy ................................................... 77 2.6.1 Introduction ........................................................................................ 77 2.6.2 Approaches to Discontinuity Modelling ................................................... 78 2.6.3 The Development of the JKMRC 3-D Discontinuity Model ......................... 85 2.6.4 The JKMRC Hierarchical Model of Discontinuity Network Geometry ........... 92 ROCK MASS CLASSIFICATION SCHEMES ................................................... 100 2.7.1 Introduction ...................................................................................... 100 2.7.2 RMR System (Bieniawski, 1974,1976) .................................................. 101 2.7.3 Q System (Barton et a/1974) ............................................................... 105 2.7.4 Modified Basic RMR or MBR System (Kendorski et a/ 1983) .................... 108 2.7.5 MRMR System (Laubscher 1990) ......................................................... 109 2.7.6 In situ Rock Mass Rating or IRMR (Laubscher and lakubec 2001) .............. III 2.7.7 Geological Strength Index (GS1) ........................................................... 114 2.7.8 Conclusions ...................................................................................... 116 THE MECHANICAL PROPERTIES OF ROCK MASSES .................................... 117 2.8.l Scope .............................................................................................. 117 2.8.2 The Hoek-Brown Empirical Strength Criterion ........................................ 117 2.8.3 Rock Mass Deformation Modulus......................................................... 122 IN SITU STRESSES ...................................................................................... 123 ix

CHAPTER 3 CAVABILITY ASSESSMENT

3.1 3.2

3.3

3.4 3.5

3.6

INTRODUCTION ......................................................................................... 126 LAUBSCHER'S CAVING CHART .................................................................. 127 3.2.1 Overvie'v.......................................................................................... 127 3.2.2 The Mining Rock Mass Rating ............................................................. 128 3.2.3 Delineation of Zones of Stability ........................................................... 129 Summary .......................................................................................... 130 3.2.4 MATHEWS' STABILITY GRAPH APPROACH ................................................ 130 3.3.1 Overview .......................................................................................... 130 3.3.2 Extension of the Method ...................................................................... 133 3.3.3 Application of Mathews' Method to the Prediction of Cavability ................. 136 NUMERICAL MODELLING APPROACHES .................................................... 138 AXISYMMETRIC CONTINUUM MODEL ....................................................... 139 3.5.1 Model Formulation ............................................................................. 139 3.5.2 Material Parameters ............................................................................ 143 3.5.3 Results ............................................................................................. 144 PFC3D DISCONTINUUM MODEL ................................................................. 147 3.6.1 Introduction ...................................................................................... 147 3.6.2 Model Description .............................................................................. 148 3.6.3 Results of Model Observations ............................................................. 151 3.6.4 Future PFC Modelling of Cavability ...................................................... 154

CHAPTER 4 FRAGME~TATION

4.1 4.2 4.3

4.4 4.5

ASSESSMENT

INTRODUCTION ......................................................................................... 156 FACTORS INFLUENCING FRAGMENTATION ............................................... !57 FRAGMENTATION MEASUREMENT ............................................................ 159 4.3.1 Overview.......................................................................................... 159 4.3.2 Digital Image Processing Methods ........................................................ 161 4.3.3 Examples of DIP Systems .................................................................... 162 4.3.4 Validation Studies .............................................................................. 165 4.3.5 Application of DIP Systems to Caving ................................................... 166 IN SITU FRAGMENTATION ASSESSMENT .................................................... 169 BCF: A PROGRAl\tl TO PREDICT BLOCK CAVE FRAGMENTATION ............... 172 4.5.1 Modelling Approach ........................................................................... 172 4.5.2 Primary Fragmentation ........................................................................ 173 4.5.3 Secondary Fragmentation .................................................................... 175 4.5.4 Hangup Analysis ................................................................................ 178 4.5.5 Discussion ........................................................................................ 179

x

4.6

4.7

AN ALTERNATIVE METHOD OF ASSESSING IN SITU AND PRIMARY FRAGMENTATION ...................................................................... 4.6.1 Methodology .................................................................................... 4.6.2 Tessellation Procedure ........................................................................ 4.6.3 In situ Blocks .................................................................................... 4.6.4 Primary Fragmentation ....................................................................... CONCLUSIONS .....................................................................................

181 181 183 186 187 190

CHAPTERS CAVE INITIATION BY UNDERCUTTING 5.1 5.2

5.3

5.4

5.5

INTRODUCTION ......................................................................................... 191 UNDERCUTTING STRATEGIES ................................................................... 192 5.2.1 Purpose ............................................................................................ 192 5.2.2 Post-Undercutting .............................................................................. 192 5.2.3 Pre-Undercutting ............................................................................... 193 5.2.4 Advance Undercutting ........................................................................ 194 5.2.5 The Henderson Strategy ...................................................................... 195 UNDERCUT DESIGN AND MANAGEMENT .................................................. 196 5.3.1 Purpose ............................................................................................ 196 5.3.2 Initiation and Direction of Undercut Advance.......................................... 196 5.3.3 Shape of the Undercut Face ................................................................. 199 5.3.4 Rate of Undercut Advance ................................................................... 200 5.3.5 Undercut Height ................................................................................ 202 UNDERCUT SHAPE AND EXTRACTION METHOD........................................ 204 5.4.1 Introduction ...................................................................................... 204 5.4.2 Fan Undercut .................................................................................... 205 5.4.3 Flat Undercut .................................................................................... 206 5.4.4 Narrow Inclined UndercuL .................................................................. 210 STRESSES INDUCED IN THE UNDERCUT AND EXTRACTION LEVELS ......... 212 5.5.1 Introduction ...................................................................................... 212 5.5.2 Modelling Strategy............................................................................. 215 5.5.3 Extraction Level Stresses Post-Undercut Sequence ................................ 217 5.5.4 Extraction Level Stresses - Advance Undercut Sequence .......................... 217 5.5.5 Undercut Level Stresses ...................................................................... 221 5.5.6 Summary of Parametric Study Results ................................................... 223 5.5.7 Undercut Drift Support and Reinforcement.. ........................................... 224

xi

5.6

DRILLING AND BLASTING FOR UNDERCUTTING AND DRAWBELL CONSTRUCTION ............................................................................................. 226 5.6.1 Introduction .......................................................................................... 226 5.6.2 Factors affecting Drilling and Blasting Performance ..................................... 227 5.6.3 Experienced based Design "Rules of Thumb" for Rock Breakage ControL ....... 229 5.6.4 Undercut Drilling and Blasting ................................................................. 234 5.6.5 Drawbell Blasting .................................................................................. 241 5.6.6 Drilling Equipment Selection ................................................................... 243

CHAPTER 6 EXTRACTION LEVEL DESIGN 6.1 6.2 6.3

6.4

6.5

PURPOSE ........................................................................................................ 245 FACTORS INFLUENCING EXTRACTION LEVEL DESIGN A"'ID PERFORMANCE .............................................................................................. 246 EXTRACTION LEVEL LAyOUTS ...................................................................... 248 Scope ................................................................................................... 248 6.3.1 6.3.2 Continuous Trough or Trench Layout ........................................................ 248 6.3.3 Herringbone Layout. ............................................................................... 250 6.3.4 Offset Herringbone Layout ...................................................................... 250 6.3.5 Henderson or Z Layout ........................................................................... 252 6.3.6 El Teniente Layout ................................................................................. 252 Ore Crushing and Transportation ............................................................... 253 6.3.7 DRA WPOINT AND DRA WBELL DESIGN .......................................................... 255 6.4.1 Gravity Flow of Caved Ore ...................................................................... 255 6.4.2 Drawpoint Spacing ................................................................................. 259 6.4.3 Drawpoint Size, Shape and Orientation ...................................................... 266 6.4.4 Dra\vbell Geometry ................................................................................ 268 SUPPORT AND REINFORCEMENT ................................................................... 270 6.5.1 Terminology ......................................................................................... 270 6.5.2 Principles ............................................................................................. 272 Support and Reinforcement Elements ......................................................... 274 6.5.3 6.5.4 Stress-Strength Analyses ......................................................................... 275 6.5.5 Support and Reinforcement of Drawpoints .................................................. 279 6.5.6 Examples .............................................................................................. 281

xii

CHAPTER 7 DRAW CONTROL

7.1 7.2

7.3 7.4

7.5

INTRODUCTION .............................................................................................. 293 DRAW MECHANISMS ...................................................................................... 295 7.2.1 Basic Studies ......................................................................................... 295 7.2.2 Mass Flow............................................................................................. 296 7.2.3 Granular or Gravity Flow ......................................................................... 296 7.2.4 Void Diffusion ....................................................................................... 297 7.2.5 Practical Implications .............................................................................. 299 DRAW CONTROL DURING UNDERCUTTING AND CAVE INITIATION ............... 30 I DRAW CONTROL DURING PRODUCTION ........................................................ 305 7.4.1 Manual Calculation of Draw Tonnages and Estimation ofDilution .................. .305 7.4.2 Draw Control Strategies and Procedures ...................................................... 309 EXAMPLES OF COMPUTERISED DRAW CONTROL SYSTEMS ........................... 311 7.5.1 PC-BC .................................................................................................. 311 7.5.2 De Beers' Linear Programming Based System .............................................. 318

CHAPTERS GEOTECHNICAL MONITORING

8.1 8.2

8.3

8.4

8.5

THE PURPOSES OF MONITORING .................................................................... 322 GEOTECHNICAL MONITORING SySTEMS ....................................................... 324 8.2.1 General Considerations ............................................................................ 324 8.2.2 What is Monitored? ................................................................................. 324 8.2.3 How is it Monitored? ............................................................................... 326 8.2.4 Where and When is it Monitored? .............................................................. 328 MONITORING THE INITIATION AND DEVELOPMENT OF CAVING ................... 331 8.3.] Why? ................................................................................................... 331 8.3.2 What and How? ..................................................................................... .33] EXTRACTION LEVEL AND INFRASTRUCTURE MONITORING .......................... 337 8.4.1 Why? .................................................................................................. .337 8.4.2 What and How? ..................................................................................... 337 8.4.3 Examples ............................................................................................. .338 MONITORING SUBSIDENCE AND GROUND MOVEMENT ................................. 342 8.5.1 Why? ................................................................................................... 342 8.5.2 What and How? ..................................................................................... 344 8.5.3 Examples .............................................................................................. 344

xiii

CHAPTER 9 SURFACE SUBSIDENCE 9.1 9.2

9.3

INTRODUCTION ............................................................................................. 346 TYPES AND MECHANISMS OF DISCONTINUOUS SUBSIDENCE ....................... 347 9.2.1 Types of Discontinuous Subsidence ........................................................... 347 9.2.2 Chimney Caving Mechanisms .................................................................. 349 EXAMPLES OF SURFACE SUBSIDENCE ARISING FROM BLOCK AND PANEL CAVING ...................................................................................... 352 9.3.1 Miami Mine, Arizona, USA ..................................................................... 352 San Manuel Mine, Arizona, USA .............................................................. 352 9.3.2

9.4

9.5 9.6

9.3.3 Henderson Mine, Colorado, USA355 ANALYSIS OF CHIMNEY CAVING AND PLUG SUBSIDENCE ............................ 356 9.4.1 Limiting Equilibrium Analysis ................................. '" .............................. 356 9.4.2 Empirical Methods ................................................................................. 362 LIMITING EQUILIBRIUM ANALYSIS OF PROGRESSIVE HANGINGWALL CAVING ................................................................................ 364 SUBSIDENCE PREDICTION IN PRACTICE ........................................................ 369 9.6.1 General Approach .................................................................................. 369 9.6.2 Prediction of Caving Induced Subsidence at Rio Blanco and El Teniente Mines, Chile ......................................................................... 370

CHAPTER 10 MAJOR OPERATIONAL HAZARDS 10.1 10.2

10.3

10.4

SCOPE ........................................................................................................... 375 MAJOR COLLAPSES ........................................................................................ 377 10.2.1 Terminology ......................................................................................... 377 10.2.2 Causes ................................................................................................. 377 10.2.3 Effects ................................................................................................ 380 10.2.4 Prevention and Amelioration .................................................................... 381 ROCKBURSTS ................................................................................................. 382 10.3.1 Terminology ......................................................................................... 382 10.3.2 Causes ................................................................................................. 382 10.3.3 Effects ................................................................................................. 384 10.3.4 Prevention and Amelioration .................................................................... 386 MUD RUSHES ................................................................................................. 387 10.4.1 Terminology ......................................................................................... 387 10.4.2 Causes ................................................................................................. 388 10.4.3 Effects ................................................................................................. 393 10.4.4 Prevention and Amelioration .................................................................... 393

xiv

10.S

10.6

AIRBLASTS ................................................................................................ 396 10.S.1 Tenninology ..................................................................................... 396 10.5.2 Causes ............................................................................................. 396 10.5.3 Effects ............................................................................................. 398 IO.S.4 Prevention and Amelioration................................................................ 398 WATER AND SLURRY INRUSHES ............................................................... 399

CHAPTER 11 RISK ANALYSIS FOR BLOCK CAVING 11.1 11.2 11.3 11.4

11.5

11.6

INTRODUCTION TO RISK ANALySIS .......................................................... 400 DEFINITIONS .............................................................................................. 40 I PROJECT DEVELOPMENT ........................................................................... 402 RISK ANALYSIS TOOLS AND CONCEPTS .................................................... 404 11.4.1 Risk Analysis Tools ........................................................................... 404 11.4.2 Sources of Risk ................................................................................. 404 11.4.3 Uncertainty....................................................................................... 40S CAVERISK .................................................................................................. 406 11.5.1 Purpose ............................................................................................ 406 11.5.2 Outline of CaveRisk ........................................................................... 406 11.5.3 Topics and Focus Issues ...................................................................... 408 I1.S.4 Likelihood and Consequences .............................................................. 413 11.5.5 Risk Detennination and Risk Acceptance ............................................... 416 11.5.6 Risk Manageability ............................................................................ 417 11.5.7 Risk Presentation .................................................................... " ......... 419 11.5.8 Rules Operating in CaveRisk ............................................................... 419 CONCLUSION ............................................................................................. 421

REFERENCES ........................................................................................................ 423 APPENDIX A: GLOSSARy ..................................................................................... 463 APPENDIX B: RELATION BETWEEN CAVED COLUMN HEIGHT AND VERTICAL STRESS AT THE CAVE BASE ......................................... 471 APPENDIX C: NUMERICAL SIMULATION OF PARTICLE FLOW USING REBOP ...... 484 APPENDIX D: LIMITING EQUILIBRIUM ANALYSIS OF PROGRESSIVE HANGINGWALL CAVING ............................................................... 501 INDEX .................................................................................................................. 509

xv

CHAPTER 1 INTRODUCTION

1.1

UNDERGROUND MINING METHODS

1.1.1

General Features

T

he underground mining of minerals involves three general sets of activities:

• • •

the development of physical access to the mineralised zone; the extraction of the ore from the enclosing rock mass; and the transport of the ore to processing facilities on the mine surface.

This general process requires the development of three main types of underground excavation: • permanent access and service openings or components of the mine infrastructure; • stope access and service openings, or stope development; and • ore sources or stopes through which the ore is removed from its in situ setting. The set of stopes generated during ore extraction by underground mass mining methods usually constitute the largest excavations formed during the overall mining process. This means that their zones of influence are relatively large compared to those of virtually all other mine openings (Brady and Brown 1993). The method by which the stopes are supported in order to maintain their fitness for purpose then becomes a major consideration in mining method selection and mine design. Indeed, it is usually on the basis of whether or not stopes are supported, and if so how, that underground mining methods are classified (eg Hamrin 1982). 1.1.2

Classification of Underground Mining Methods

Most systems of classifying underground mining methods are based on methods of supporting the stopes. As Rossouw and Fourie (1996) have argued, the classification of underground mining methods is not as straightforward as might be supposed. In order to overcome some of the perceived difficulties with existing systems, they proposed a three-dimensional presentation which takes into account three forms of support - natural (pillars), artificial (fill) and none

1

Chapter 1: Introduction

(caving). However, Roussow and Fourie’s presentation is quite complex and has not found widespread use. The essential features to be considered are the relations between the method of working, the key orebody properties defining the applicability of that method and the country rock mass properties that are essential to sustain the method (Brady and Brown 1993). Figure 1.1 shows one version of a common approach to underground mining method classification. Not all methods of mining currently employed are shown on this diagram (eg bench stoping) but they could be added if required. The unsupported or caving methods of mining seek to induce mass failure of, and large displacements in, the country rock which will necessarily behave as a discontinuum. At the other end of the spectrum, the supported methods seek to maintain the integrity and “elastic” response of the country rock and to strictly limit its displacement.

Figure 1.1: Classification of underground mining methods (Brady and Brown 1993)

As shown in Figure 1.1, the unsupported or caving methods of mining include block (and panel) caving, sublevel caving and longwall methods. In the longwall method applied to coal mining, the mineral (coal) is extracted mechanically and the overlying strata cave under the influence of redistributed stresses and gravitational forces. The longwall methods used to mine the deep, flat dipping, tabular gold reefs in South Africa are sometimes classified as caving methods (eg Brady and Brown 1993), although the mechanism by which the overlying rock displaces to fill the void created by the extraction of the ore usually involves “elastic” displacement of the rock on the release of extremely high stresses rather than, or as well as, caving per se. In sublevel caving methods, the ore is drilled and blasted and drawn following which the surrounding waste rock caves naturally. In the block and panel caving methods with which this book is concerned, both the ore and the overlying rock cave under the influence of

2

Chapter 1: Introduction

gravity and the redistributed in situ stresses once the orebody has been undercut. In these methods, in particular, the caving and caved ore and waste rock behave as discontinuous materials. 1.2 1.2.1

BLOCK AND PANEL CAVING Outline of the Method

Figure 1.2 illustrates the general features of the block caving method. In this method, the full orebody or an approximately equidimensional block of ore is undercut fully to initiate caving. The undercut zone is drilled and blasted progressively and some broken ore is drawn off to create a void into which initial caving of the overlying ore can take place. As more broken ore is drawn progressively following cave initiation, the cave propagates upwards through the orebody or block until the overlying rock also caves and surface subsidence occurs. The mechanisms by which caving takes place under the influence of redistributed stresses and/or gravity will be outlined in Section 1.2.2.

Figure 1.2: Example of block caving with LHD loaders, El Teniente, Chile (Hamrin 2001)

3

Chapter 1: Introduction

The broken ore is removed through the production or extraction level developed below the undercut level and connected to it by drawbells through which the ore gravitates to drawpoints on the extraction level. In most current block caving operations, the broken ore is removed from the drawpoints by Load-Haul-Dump (LHD) vehicles although some still use the more traditional gravity - based grizzly or slusher systems as discussed in Section 1.2.3. From the extraction level, the ore is transported to the haulage level and out of the mine, sometimes following underground crushing. Block caving may be used in massive orebodies which have large, regular “footprints” and either dip steeply or are of large vertical extent. It is a low cost mining method which is capable of automation to produce an underground “rock factory” (eg Tota 1997). However, it is capital intensive requiring considerable investment in infrastructure and development before production can commence. Historically, block caving was used for massive, low strength and usually low grade orebodies which produced fine fragmentation (Lewis and Clark 1964). Where mining is to be mechanised, the low strength of the rock mass can place limitations on the practicable sizes of the extraction level excavations. Furthermore, finely fragmented ore can “chimney” when drawn requiring the drawbells to be closely spaced so that undrawn “pillars” of broken ore do not form (Ward 1981). These factors place limitations on the sizes of the equipment that can be used. Accordingly, there is now a tendency for the method to be used in stronger orebodies which produce coarser fragmentation than did the traditional applications of the method. This enables more widely spaced drawpoints and larger equipment to be used. Panel caving and other variants of the general method such as inclined drawpoint caving and front caving, operate on the same principles as block caving. In panel caving, the orebody or mining block is not undercut fully initially but, rather, a panel or strip of the orebody is undercut and allowed to cave. Development, undercutting and mining of the subsequent panels then follow some distance behind the first panel as illustrated in Figure 1.3. As a result, the cave front moves across the block or orebody at a constant angle to the direction of advance of the undercut. Examples of the application of this method will be given in Section 1.3 below. Inclined drawpoint caving (Laubscher 2000, Laubscher and Esterhuizen 1994) is used when it is not possible to develop the drawpoints on one level, usually because the orebody has a welldefined inclined footwall. In this case, the drawpoints are developed at the footwall contact from the footwall on successive sublevels with the drifts being continued to serve as undercut drill drifts. In some cases such as that at the King Mine, Zimbabwe, illustrated in Figure 1.4, local geological conditions may lead to a “false footwall layout” being used in which the inclination of the plane of the drawpoints is flatter than the footwall contact (Laubscher 2000, Laubscher and Esterhuizen 1994). Front caving was developed from the overdraw system used on the two lower levels of the sublevel caving operations at the Shabanie Mine, Zimbabwe (Laubscher 2000). In recent years, front caving has been used at the Koffiefontein Mine, South Africa, and the King Mine,

4

Chapter 1: Introduction

Zimbabwe, where the method is referred to as retreating brow caving. In essence, the method involves retreating on one or more levels from an initiating slot which can be in the centre of the orebody as at Koffiefontein, or against the orebody boundary. The lower level is the production level on which so-called semi-permanent drawpoints are fully developed ahead of undercutting on the upper level. This upper level also provides initial temporary drawpoints from which the swell from each blasted ring is drawn. The undercut is retreated in stages to points above the semi-permanent drawpoints in a manner similar to that used in sublevel caving. Ideally, the method should work best with two production levels rather than one. However, this approach may be precluded on cost or other grounds, including space and layout considerations.

Figure 1.3: Mechanised panel caving, Henderson Mine, Colorado, USA (Doepken 1982)

There are many more variants of block and panel caving methods of mining than those listed above. For example, the macrotrench (or macrozanja) method developed at the El Teniente Mine, Chile, contains elements of panel, inclined drawpoint and front caving methods. Exploitation starts through a four level sublevel cave that begins from a central slot and is then retreated to both sides leaving a large trench around the initial slot. The sublevel caving is stopped in a position which leaves the upper levels and their drawpoints more advanced than the lower levels (Diaz and Tobar 2000).

5

Chapter 1: Introduction

(a)

(b) Figure 1.4: Inclined drawpoint caving, King Mine, Zimbabwe, (a) vertical section showing extraction level layout, and (b) plan showing sublevel layout (Laubscher and Esterhuizen 1994)

6

Chapter 1: Introduction

Caving methods of mining may be classified according to • • • •

whether or not part of the ore column is broken by blasting or other "artificial" methods; whether or not a crown pillar is left between mining lifts; the undercutting strategy used (see Section 1.3 and Chapter 5); and the method of ore loading used.

Figure 1.5 shows an informative classification of caving methods of mining, including sublevel caving, developed on this basis by Flores and Karzulovic (2002).

Figure 1.5: Classification of caving methods of mining (Flores and Karzulovic 2002) Many of the larger orebodies being mined by the caving method in fact use panel caving although the more generic term block caving may sometimes be used to describe the mining method. Generally in this book, the term block caving will be used as a generic rather than as a specific term so that the discussion will usually apply to panel caving as well.

7

Chapter 1: Introduction

Mine

It will be apparent from this introductory description of block and panel caving methods, that while their capital or development costs may be relatively high, operating costs can be expected to be lower than those of other underground mining methods. It is for this reason that caving methods are attractive for the mass mining or large, lower grade orebodies. Figure 1.6 summarises the underground mining cash costs in $US per tonne at a number of block and panel caving operations in the years 1999 and 2000. (These data were compiled by Northparkes Mines, Australia, and shared with ICS sponsors).

Underground mining cash cost US$/tonne Figure 1.6: Comparative underground mining cash costs for block and panel caving mines in 1999 and 2000 1.2.2

Basic Caving Mechanics

It must be expected that any unsupported rock mass will cave if it is undercut to a sufficient extent. As has been noted earlier, caving occurs as a result of two major influences – gravity and the stresses induced in the crown or back of the undercut or cave. The mechanisms by which caving occurs will depend on the relationships between the induced stresses, the strength of the rock mass and the geometry and strengths of the discontinuities in the rock mass. Much accumulated experience supports the contention of Kendorski (1978) that the successful initiation and propagation of caving requires the presence of a well-developed, low-dip discontinuity set. The structure most favourable for caving has been found to be one in which a low-dip discontinuity set is augmented by two steeply dipping sets which provide conditions suitable for the vertical displacement of blocks of rock (eg Mahtab et al 1973).

8

Chapter 1: Introduction

If the compressive tangential stresses induced in the crown of the undercut or cave are low, or tend to be tensile, blocks of rock may become free to fall under the influence of gravity or to slide on inclined discontinuities. These conditions may occur when the horizontal in situ stresses are low or where boundary slots or previous mining have relieved the stresses or redistributed them away from the block or panel being mined. Even under these circumstances, it is sometimes possible for a self-supporting arch to develop in the crown of the cave, especially if an appropriate draw control strategy is not in place. Some of the mechanisms by which caving and arching may occur under these low lateral stress conditions are illustrated by the simple and idealised distinct element simulation shown in Figure 1.7. Each pair of drawings in Figure 1.7 represent the geometric configuration of the blocks and the interblock contact force vectors at different stages in the progressive caving of the mass. Note that two apparently independent arches form where high levels of inter-block force traverse the mass. The upper arch is stronger and is sustained longer than the lower arch but both fail eventually by slip at the rigid abutments. At the other extreme, when the induced tangential stresses are high compared with the compressive and shear strengths of the rock mass and the shear strengths of the discontinuities, failure may occur at or near the boundary of the rock mass and blocks or slabs of rock may become free to fall under the influence of gravity. Under these circumstances, the dominant mechanisms of failure are brittle fracture of the intact rock and slip on discontinuities, especially those that are flat dipping (eg Heslop and Laubscher 1981). This form of caving is sometimes referred to as stress caving. Duplancic and Brady (1999) used a seismic monitoring system to study the early stages of caving of Lift 1 at Northparkes Mines’ E26 block cave, New South Wales, Australia. From the data collected and analysed, they developed the conceptual model of caving for this case shown in Figure 1.8. The model contains five regions described by Duplancic and Brady (1999) in the following terms: 1.

Caved zone. This region consists of rock blocks which have fallen from the cave back. Material in the caved zone provides support to the walls of the cave.

2.

Air gap. During continuous caving, the height of the air gap formed is a function of the extraction rate of the material from the caved zone.

9

Chapter 1: Introduction

Figure 1.7: Idealised distinct element simulation of block caving (after Voegele et al 1978)

3.

Zone of discontinuous deformation. This region no longer provides support to the overlying rock mass. Large-scale displacements of rock occur in this area, which is where disintegration of the rock mass occurs. No seismicity is recorded from within this region. The zone was estimated to extend 15 m from the boundary of the cave crown.

10

Chapter 1: Introduction

4.

Seismogenic zone. An active seismic front occurs due to slip on joints and brittle failure of rock. This behaviour is due to changing stress conditions caused by the advancing undercut and progress of the cave.

5.

Surrounding rock mass. Elastic deformation occurs in the rock mass ahead of the seismic front and surrounding the cave. Pseudo-continuous domain Seismogenic zone

Zone of loosening Air gap

Caved zone

Direction of advancing undercut Figure 1.8: Conceptual model of caving (Duplancic and Brady 1999) Duplancic and Brady’s observations at Northparkes confirm the previous general finding that for boundary collapse to occur, a flat lying discontinuity set is required to act as a release mechanism. A third general case must be considered. If the horizontal in situ stresses and the tangential stresses induced in the crown of the undercut or cave are high enough to develop clamping forces which inhibit gravity-induced caving, but are not high compared with the compressive strength of the rock mass, caving may be inhibited and a stable arch may develop. Under these circumstances, some form of cave induction may be required to weaken the rock mass, relieve the tangential stresses or induce slip on discontinuities (eg Kendrick 1970, van As and Jeffrey 2000). A different mechanism from those discussed so far is involved in subsidence caving in which a large mass of rock subsides rapidly as a result of shear failure on the vertical or near-vertical boundaries of a block. For this to occur the normal (horizontal) stresses developed on the vertical boundaries of the block, or the shear strength of the interface, must be so low that the total shear resistance developed is unable to resist the vertical force due to the weight of the

11

Chapter 1: Introduction

block. For such a failure to have catastrophic consequences, there would need to exist a large mined-out void into which the caving mass could fall. This circumstance would not arise in a block or panel cave if the draw control strategy used did not allow a significant air gap to develop below the cave back. Once continuous caving has been initiated, the rate of production from the block or panel will depend on the rate at which the cave propagates following draw and the creation of a small air void into which caved material may fall. In practice this rate of caving will depend on the rate of undercutting, the quality of the rock mass and the magnitude of the induced stresses. As will be discussed in Chapter 5, the direction of undercutting with respect to the in situ stress orientation is also important. Estimated caving rates for a number of mines are summarised Table 1.1. It should be emphasised that these caving rates are estimated. They are notoriously difficult to measure. Furthermore, they may vary through the life of a cave. For example, as the height of the cave and of the column of broken ore increases, the induced stresses in the cave back may change, as may the structure and rock mass strength of the orebody. Table 1.1: Estimated caving rates Operation

Estimated Caving Rate (mm per day)

CODELCO El Teniente Sub 6 panel cave

200 to 300

CODELCO Esmeralda panel cave

170 to 200

De Beers Koffiefontein (TKB Kimberlite)

200 to 400

De Beers Premier Mine (TKB Kimberlite)

100 to 1200

De Beers Premier Mine (HYB Kimberlite)

60 to 250

Henderson Mine

270

Northparkes E26 Lift 1 block cave

110 to 380 (pre inducement)

Under steady-state production conditions, the average rate of draw will be a function of the rate of natural caving and the bulking factor of the caved ore. In currently operating block and panel caving mines, rates of draw vary up to about 700 mm/day with the mean in the range 200 to 250 mm/day (Flores and Karzulovic 2002b). Drawing of the difference between the in situ and caved volumes following each caving episode will ensure that cave propagation is controlled and an excessive air gap does not develop. Of course, for this controlled caving to occur, a small air gap must be created by drawing the caved ore. The major consequence of allowing an excessive air gap to develop is the danger of massive rock falls and the associated air blasts to be discussed in Chapter 10.

12

Chapter 1: Introduction

1.2.3

History of Block Caving

The precursor of the modern block caving method of mining was developed in the iron ore mines of the Menominee Ranges, Michigan, USA, in the late nineteenth century. The Pewabic Mine was the first to use a form of block caving from which other methods developed (Peele 1941). In the Pewabic method, blocks of ore approximately 60-75 m long, 30-40 m high and the full width of the deposit (60 m) were caved in one operation. An unusual feature of the method from a modern perspective is the fact that the ore was handled by shovelling in drifts driven and kept open within the caved mass (Peele 1941). Variations of the Pewabic method were soon developed at other iron ore mines in Michigan and, from the early part of the twentieth century, in the copper mines in western USA. Before evolving to the use of full block caving, many mines initially used combined methods involving, for example, shrinkage stoping and caving methods for the subsequent mining of the pillars between the primary stopes. Peele (1941) gives several examples of these combined methods. A good example of the early application of a full block caving method is provided by the Miami Copper Company’s mine in Arizona. Descriptions provided by Peele (1941) and Lewis and Clark (1964), and the diagram shown in Figure 1.9, are based on a paper by McLennan (1930). (Note that the original dimensions in feet have been retained in Figure 1.9. They have been approximated in metres in the text). The flat lying orebody of considerable lateral extent varied in thickness to more than 60 m and was over- and under-lain by waste. Early mining was by top-slicing but this was replaced by shrinkage stoping with sublevel caving of the pillars and, in the 1920s, by block caving. Initially, caving practice involved undercutting and caving the orebody across its entire width of 150-200 m, starting at one end and retreating along the length of the orebody. This approach was unsuccessful and later practice was to cave and draw alternate 45 m wide panels across the entire orebody. When the waste rock had settled into the original panels and compacted, the pillar panels were caved. This method was satisfactory for moderate thicknesses of ore averaging 60 m but was modified to true block caving where thicknesses were 90 m or more. The original caving blocks of the thicker ore were 45 by 90 m in plan as shown in Figure 1.9. Experience showed that 45 m square blocks gave better results and this block size became the standard. The order of mining was such that a block being mined would be entirely surrounded by either solid ore or mined-out blocks in which the capping had settled until it was quite compact (Lewis and Clark 1964). As shown in Figure 1.9, the Miami mine used a gravity system of ore transfer to the haulage level incorporating grizzlies. By the 1920s and 30s, block caving methods were being used in a wide range of mines exploiting massive, weak orebodies. During this period, the method was introduced, for example, at the King Mine which mined asbestos in Quebec, Canada, the Climax Mine mining molybdenum in Colorado, USA, and the copper mines in Chile (Peele 1941). 13

Chapter 1: Introduction

Figure 1.9: Block caving, Miami Mine, Arizona, USA (Lewis and Clark 1964) South Africa’s first kimberlite diamond pipe was discovered at Kimberley in 1870 a few years after the discovery of alluvial diamonds at Hopetown. The early mining of this and other kimberlite pipes was by surface methods. An initial attempt at underground mining from 1884 was unsuccessful, largely because of the uncooperative attitudes of the many claim holders (Owen and Guest 1994). The consolidation of the pre-existing companies into De Beers Consolidated Mines in 1888 provided the impetus for the successful introduction of underground mining at Kimberley from 1890. A complex mining method known as chambering was used until it was replaced progressively from the late 1950s to the early 1970s (Hartley 1981). Peele (1941) describes this method as a combination of shrinkage stoping and sublevel caving. A visit to North America in the early 1950s convinced De Beers mine managers of the potential of the block caving method for mining the diamondiferous pipes (Gallagher and Loftus 1960). After an experimental block cave had been mined successfully at the Bultfontein Mine it was decided to introduce block caving on a large scale at several of the De Beers mines. The first mine to change fully from chambering to block caving was Jagersfontein where the transition was completed in 1958 (Gallagher and Loftus 1960).

14

Chapter 1: Introduction

As Owen and Guest (1994) describe, a number of different mining methods including sublevel caving, vertical crater retreat and benching have been used at the various De Beers mines but variants of block caving remain the main group of methods used currently. A feature of the evolution of the current mining methods is that not all methods introduced were successful initially and mining plans often had to be revised. Mechanised panel caving was introduced at the Premier Mine from 1990 (Bartlett 1992). Mining operations at Premier are summarised in Section 1.3.3 below. As has been indicated above, ore has been drawn throughout much of the history of block and panel cave mining by gravity or slusher methods. In 1981, Pillar (1981) listed a wide range of well-known caving mines as then using the slusher method. The availability of LHD equipment from the 1960s has provided the potential for the introduction of mechanised and trackless cave mining, especially for the more coarsely fragmenting and stronger ores in which the necessary large extraction level openings can be developed and maintained. From the 1980s, many of the major caving mines have introduced mechanised methods (see Table 1.2) although gravity systems are still used, particularly in the weaker ores. Examples of modern mechanised methods are given in Section 1.3. Table 1.2: Examples of current caving operations Mine

Country

Type

Ore Type

Annual Tonnage

Northparkes

Australia

Block

Copper-Gold

4 Mt/y

Freeport IOZ

Indonesia

Block; LHD

Copper-Gold

7 Mt/y

Palabora

South Africa

Block; LHD

Copper

10 Mt/y

CODELCO

Chile

Panel; LHD

Copper

35 Mt/y

Chile

Panel; LHD

Copper

16 Mt/y

Copper

2.5 Mt/y

E26 Lift 1

El Teniente Division CODELCO Andina Division CODELCO

Grizzly Chile

Panel; LHD

Salvador Premier Mines

South Africa

Panel; LHD

Diamonds

3 Mt/y

Henderson

USA

Panel; LHD

Molybdenum

6 Mt/y

Philex

Philippines

Grizzly; LHD

Copper

Shabanie

Zimbabwe

Retreating brow

Chrysotile asbestos

Tongkuangyu

China

Block; LHD

Copper

San Manuel

USA

Block; Grizzly

Copper

15

4 Mt/y

Chapter 1: Introduction

1.3 1.3.1

BLOCK AND PANEL CAVING OPERATIONS Overview

Because they are mass mining methods having low production costs, block and panel caving methods are currently important sources of mineral production on a world scale. Laubscher (1994) estimated that, at that time, these methods accounted for a total ore production of approximately 370,000 tonnes per day. Because of the high productivity of caving methods and the potential that they offer for mechanisation and reduced labour costs, there is a current tendency in the industrially advanced nations, in particular, to apply block caving to stronger orebodies than those to which the method has been applied in the past. This brings with it particular challenges in predicting caveability, an issue to be introduced in Section 1.4.1 and explored in detail in Chapter 3. Block caving is also being considered for the underground mining of some major orebodies previously mined by large open pits. Significant examples of this development are the Bingham Canyon mine in Utah, USA (Carter and Russell 2000) and the Palabora mine in the Republic of South Africa (Calder et al 2000). Table 1.2 lists many of the world’s major block and panel caving operations in the year 2000. The list is intended to be indicative rather than exhaustive. Because of a paucity of accessible data, cave mining operations in some countries such as China and the countries of the former USSR are not well represented in Table 1.2. Three of the major mines listed in Table 1.2 – El Teniente, Premier and Henderson - are described in the subsequent sub-sections as important but differing examples of modern cave mining practice. These descriptions may use some terms that have not been defined so far and which may not be familiar to some readers. In these cases, reference should be made to that part of the book in which the term is introduced in more detail or to the glossary of terms presented at Appendix A. Recent accounts of a number of other block and panel caving operations are given by Hustrulid and Bullock (2001). 1.3.2

El Teniente Mine, Chile

El Teniente is a division of CODELCO-Chile, Chile’s Government-owned copper mining company. The mine is located 130 km southeast of Santiago in the foothills of the Andes mountains. With production of about 100,000 tonnes per day, El Teniente is the world’s largest block or panel caving mine. The large copper porphyry orebody which was discovered in 1760, reaches a depth of more than 1 km below surface and almost completely surrounds a roughly 800-1000 m diameter circular breccia pipe known as the Braden Pipe. The orebody is approximately triangular in plan and has a radial extent of between 400 and 800 m from the perimeter of the pipe. The major items of mine infrastructure are conveniently located within the breccia pipe. Figure 1.10 shows a schematic view of the several levels in the El Teniente mine. The early block caving production level, Teniente 1, is located some 650 m above the main access level, Teniente 8, which is at an elevation of 1983 m. 16

Chapter 1: Introduction

After simple and irregular open pit mining of the near-surface secondary enriched ore, underground mining began in 1906 using overhand stoping and room and pillar methods. From 1940, a traditional gravity flow block caving method was successfully introduced (Kvapil et al 1989). The secondary ore was weak, fragmented readily and was well suited to mining by block caving. However, the quantity of the secondary ore decreases with depth so that from about 1982, mining has been increasingly in the lower grade and stronger primary ore. The primary ore is an andesite which contains pockets of even stronger diorite and dacite. Mechanised panel caving using LHDs was introduced on the Teniente 4 level (at 2347 m) in 1982 (Alvial 1992).

Figure 1.10: Schematic representation of levels, El Teniente Mine (Kvapil et al 1989) A major factor influencing the mining of the El Teniente orebody, particularly at increasing depth in the primary ore, has been the existence of extremely high lateral in situ stresses associated with the nearby subduction zone in which the Pacific plate is thrust under the edge of

17

Chapter 1: Introduction

the South American plate. Rock bursts were first experienced on the Teniente 4 level in 1976 (Alvial 1992) and have been a continuing problem since mechanised mining began in the stronger and stiffer primary ore. El Teniente now uses mainly mechanised panel caving but there is still some production from areas which use other forms of caving (Jofre et al 2000). Accounts of successive stages in the evolution of cave mining methods at El Teniente are given by Ovalle (1981), Kvapil et al (1989), Alvial (1992), Moyano and Vienne (1993), Rojas et al (2000b), Jofre et al (2000) and Rojas et al (2001). The following description of recent mining at El Teniente draws on the accounts of Moyano and Vienne (1993) and Jofre et al (2000). As has been noted, mechanised panel caving of the primary ore was introduced on the El Teniente 4 level in 1982. A conventional or post-undercutting method was used with the undercut being mined after the development of the extraction level below and the excavation of the drawpoints. (Details of the meaning and implications of post-undercutting and related terms are given in Chapter 5). In the traditional El Teniente panel caving method, the undercut level was located 18 m above the extraction level and the undercut drifts were 3.6 by 3.6 m on 30 m centres. Over time the height of the undercut has been progressively reduced from 16.6 m in 1987 to 10.6 m in 1998 (Jofre et al 2000). The extraction level layout developed at El Teniente is illustrated in Figure 1.11 for the Teniente Sub 6 level. The conventional or post-undercutting sequence used until recently exposed the pre-constructed extraction level to high levels of abutment stress as the undercut advanced overhead. This resulted in damage to the extraction level pillars and, in some instances, in severe rock bursts associated with the high horizontal stresses and local geological features (Moyano and Vienne 1993, Rojas et al 2000a). A series of major rock bursts on the Teniente Sub 6 level starting in January 1990 soon after it came into production, caused production to be stopped temporarily and the approach to mining revised. The measures introduced from 1994 to successfully control the incidence and severity of rock bursts on the Sub 6 level included reducing the height of the column of intact rock above the undercut level, ensuring an even spatial and temporal rate of extraction, mining at a slow rate especially initially, using only remotely controlled production equipment from 1995 and using the results of seismic and other monitoring to guide production planning. In 2000, the production rate of Teniente Sub 6 was about 10,000 tonnes per day and the undercutting rate was about 12,000 m2 per year (Rojas et al 2000a).

18

Chapter 1: Introduction

Notes:

1. 2.

Haulage drift and drawpoint drifts section wide 4.00 m : high 3.60 m Distance between consecutive dump points located in the same haulage drift will be that corresponding to 6 drawpoint drifts (103.92 m)

Figure 1.11: Extraction level layout, Sub 6 level, El Teniente Mine (Moyano and Vienne 1993) From 1992 a pre-undercut panel caving method was tested at El Teniente. In 1997 the method was brought into production on the new Esmeralda section on the Teniente Sub 5 level. In this method, the undercut level is developed and then blasted in advance of the development of the extraction level and formation of the drawbells. Thus, all extraction level development and construction takes place in a de-stressed zone below the mined undercut. The undercut drifts are 3.6 by 3.6 m on 15 m centres. Initial development on the extraction level is kept 22.5 m behind the undercut, and full construction on the extraction level occurs 45 to 60 m behind the undercut (see Figure 1.12). Experience and the results of extensive monitoring of the condition of the pillars and the extraction level installations, clearly demonstrate the effectiveness of this approach (Rojas et al 2000b, Jofre et al 2000). In early 2000, the average production from the Esmeralda section was 12,500 tonnes per day with a planned peak of 15,000 tonnes per day. The pre-undercut method was also being used on the Teniente 3 Isla section.

19

Chapter 1: Introduction

PRE UNDERCUT

Figure 1.12: Comparison of conventional panel caving and pre-undercut activity sequences, El Teniente Mine (after Rojas et al 2000b) 1.3.3

Premier Diamond Mine, South Africa

The Premier Diamond Mine is situated 45 km to the east-north-east of Pretoria, Republic of South Africa. It exploits the Premier Pipe, the largest of South Africa’s kimberlite pipes having a surface area of 32 hectares. The Premier Pipe is unique geologically in that it is intersected by a 75 m thick, shallow-dipping gabbro sill which cuts through the pipe at depths of between 380 and 510 m below surface (Bartlett 1992). Figure 1.13 shows a diagrammatic vertical section and plan of the geology and of the recent and planned mining blocks. Open pit mining commenced at Premier in 1902. With the increasing depth of mining, underground mining started in 1948, initially by open benching. Following the successful implementation of the block caving method at other De Beers mines at Kimberley and Jagersfontein as outlined in Section 1.2.3, block caving was introduced at Premier in the early 1970s. Four separate caves using a grizzly system feeding slusher drifts were operated above the gabbro sill on the western side of the mine. A total of 85 million tonnes of ore was produced from these caves (Bartlett 1992).

20

Chapter 1: Introduction

Figure 1.13: Diagrammatic section and plan, Premier Diamond Mine (Bartlett and Croll 2000) Mining below the sill started in 1979 using an open stoping method. This method was unsuccessful and was replaced by block caving. Cave mining using LHDs to transport the ore from the drawpoints to ore passes started in 1990 in the BA5 mining block on the 630 m level on the western side of the mine (see Figure 1.13). A second cave was established in 1996 in the 21

Chapter 1: Introduction

BB1E block at a depth of 732 m in weaker ore on the eastern side of the pipe. Possible future mining of the C-cut (see Figure 1.13) would involve the exploitation of 170 million tonnes of ore by caving methods with a production level 1000 m or more below surface (Bartlett and Croll 2000). The remainder of this Section will discuss recent and planned caving operations at Premier as reported by Bartlett and Croll (2000). Accounts of the earlier stages of cave mining at Premier are given by Owen (1981), Bartlett (1992) and Owen and Guest (1994). Mining of the BA5 block was planned as a panel retreat caving operation with a post-undercut mining sequence. With this sequence, mining of the undercut would occur after the extraction level development had been completed. Undercut drifts were developed directly above the extraction drifts on 30 m centres. The block was undercut by drilling and blasting 120 m long by 30 m wide and 20 m high slots at right angles to the directions of the undercut and extraction level drifts. As undercut rings were blasted, broken ore dropped directly into the pre-developed drawbells. As the undercut area approached that required for the onset of caving, the levels of stress on both the undercut and extraction levels increased. On the undercut level, it became increasingly difficult to drill, charge and blast the long holes and the rate of undercutting slowed. On the extraction level, shotcrete linings were extensively damaged by the abutment stresses associated with the now slow moving undercut. Footwall heave was widespread, damaging concreted roadways and disrupting production. When continuous caving was established, the stress levels stabilised but an extensive program of extraction level support and rehabilitation was required to ensure safety and uninterrupted production. The rate of caving was slower than planned and a large and potentially dangerous air gap developed. The caving process was compromised by the presence of the overlying, strong gabbro sill. The difficulties continued as the panel retreated to the east. It was then decided to adopt an advance undercut mining sequence in which only the production drift and drawpoint breakaways were developed and partly supported ahead of the mining of the undercut overhead. The height of the undercut was also reduced. As the zone below the advancing undercut became de-stressed, development of the extraction level, including the concreting of roadways and the application of a shotcrete lining, was completed. The adoption of these measures allowed undercutting to proceed at the planned rate, support and rehabilitation requirements to be reduced and production targets to be met. The lessons learned from the BA5 block were applied in planning the mining of the BB1E block (see Figure 1.13) as a panel retreat cave with an advance undercut mining sequence. Table 1.3 shows a comparison of the design and operational parameters of the BA5 and BB1E blocks. As Bartlett and Croll (2000) explain, some difficulties were experienced in the BB1E block, particularly in the early stages of undercutting and cave establishment. However, once the cave was fully established and the undercut could be advanced at the planned rate, satisfactory results were achieved. Experience with the BB1E block shows that successful 22

Chapter 1: Introduction

caving at this depth requires detailed planning, the timely availability of resources, and careful implementation and control of the entire mining process. Maintaining the planned schedule to avoid the compaction of broken ore provided a particular challenge in this case.

Table 1.3: Design and operational parameters, BA5 and BB1E mining blocks, Premier Mine, South Africa (Bartlett and Croll 2000) Parameter Column height Rock mass rating Hydraulic radius Mining sequence Rate of undercutting Tons in mining block Tons per drawpoint Drawpoint spacing Distance across major apex Average rate of draw Initial fragmentation Fragmentation after 20 % drawn Drawpoint support Brow wear Tunnel size Lhd type Tons per LHD per hour LHD average tramming distance Hangup frequency Fragmentation Initial 20% drawn Secondary blasting

BA5 Mining Block 80 - 140 metres 45 - 65 30 Post and advance undercut 900 square metres per month 42 million tons 50 000 to 120 000 tons 15 x 15 metres 22.6 metres 180 mm per day (109 tons) 30 % >2 cubic metres 12 %>2 cubic metres

BBIE Mining Block 148 - 163 metres 45 - 55 25 Advance undercut 1100 square metres per month 23 million tons 100 000 to 200 000 tons 15 x 18 metres 23 metres 165 mm per day (120 tons) 30 % >2 cubic metres 7 % >2 cubic metres

Cable anchors, rockbolts, mesh tendon straps and shotcrete 0 to 2 metres wear after 50 000 tons drawn 4 x 4.2 metres Diesel and electric 5 and 7 yard Toros 118 tons 154 metres

Cable anchors, rockbolts, mesh tendon straps and shotcrete 1 to 3 metres wear after 50 000 tons drawn 4 x 4.2 metres Diesel and electric 5 and 7 yard Toros 131 tons 134 metres

30 % of drawpoints per shift

25 % of drawpoints per shift

30 % >2 cubic metres 12 % >2 cubic metres 40 grams per ton

30 % >2 cubic metres 10 % >2 cubic metres 30 grams per ton

The C-cut mining block is being planned with a pre-undercut mining sequence to avoid extensive and possibly unsustainable damage to the extraction level at a depth of more than 1000 m. As a consequence, the mining sequence in the existing BB1E mining block is being changed to gain operational experience with this method in which the undercut is fully developed before the extraction level. The height of the cave is planned to be between 350 and 450 m compared with a maximum of 164 m in BB1E (see Table 1.3). New mine infrastructure is being installed to support mining of the C-cut, including two new shafts from surface and a new processing plant. The mine is being designed to extract and process 9 million tonnes of ore annually by 2008 (Bartlett and Croll 2000).

23

Chapter 1: Introduction

1.3.4

Henderson Mine, Colorado, USA

The Henderson mine is located 80 km to the west of Denver, Colorado, USA, and 3170 m above sea level on the eastern side of the Continental Divide. The top of the molybdenite orebody lies more than 1000 m below the peak of Red Mountain and the lowest excavation is at a depth of 1600 m making Henderson one of the deepest caving operations in the world. The deposit is elliptical in plan with axes of 670 and 910 m. As is illustrated in Figure 1.14, the ore is transported by conveyor from a crusher complex on the 7065 level at an elevation of 2153 m to the mill site 25 km away on the western side of the Continental Divide. This discussion of the Henderson Mine is drawn from the paper by Rech et al (2000). Earlier accounts of aspects of the operation are provided by Brumleve and Maier (1981), Doepken (1982) and Rech et al (1992).

Figure 1.14: General section, Henderson Mine (after Rech et al 2000) Henderson commenced operation in 1976 as a mechanised panel cave with rail haulage from the 7500 level at an elevation of 2286 m. Figure 1.15 illustrates the original panel caving method. Approximately 90 million tonnes of ore were produced from the 8100 level (at 2469 m) from 1976 to 1991. In 1992, the 7700 level (at 2347 m) was brought into production and by the year 2000 had produced more than 45 million tonnes of ore. A particular feature of the operation of the Henderson mine is that during its life, the world molybdenum market has experienced several phases of over- and under-supply with the result that the mine’s production rates have had to vary accordingly. This has required flexible planning and operation of the mine.

24

Chapter 1: Introduction

The next production level to come into operation will be the 7225 level located 145 m below the 7700 level at an elevation of 2202 m. As is shown in Figure 1.14, the eastern section of the orebody has not been exploited from either the 8100 or the 7700 levels. Columns of ore up to 244 m high will become accessible on the eastern side from the 7225 level (Rech et al 2000). Figure 1.15 shows a typical isometric section of the recent configuration of the operating section of the mine on the 7700 or 2347 m level. The undercut level at 2364 m is developed with 3.7 by 3.7 m drifts on 24.4 m centres. Future undercut drift spacing will be 30.5 m centres. From the undercut drifts, rings of both short and long holes are drilled on 2 m centres to mine the undercut and the drawbells simultaneously. Panels are 8 to 12 production drifts (244 to 366 m) in width.

Figure 1.15: General isometric view, mechanised panel caving, Henderson Mine (Rech et al 2000)

25

Chapter 1: Introduction

Figure 1.16 shows a plan view of the drawpoint and production drift layout used on the extraction level located 17 m below the undercut level. The entry angles of 56o are the sharpest that can be used with the current 7 m3 LHDs. Drawpoints are concrete lined and are fitted with steel wear plates to protect the openings from degradation over their production lives. The roadways and drawpoint floors are lined with a 300 mm thickness of concrete which permits effective clean-up and reduces tyre wear. There is a ventilation level at 2333 m through which intake and exhaust air is transported on separate horizons. The truck haulage level at 2153 m consists of 6 by 6 m haulage drifts that provide access of the 72.6 tonne side dumping trucks to centre loading chutes at the bottoms of the ore passes. The trucks transport the ore to the gyratory crusher dump on this level.

20.57 m x 30.48 m Drawpoint spacing

Figure 1.16: Extraction level plan, Henderson Mine (Rech et al 2000) The sequence in which the development of this series of openings takes place must be well planned and coordinated. Figure 1.17 shows the general two-year sequence in the development of a panel at Henderson.

26

Chapter 1: Introduction

Figure 1.17: Development sequence, Henderson Mine (Rech et al 2000)

1.4

RISK IN CAVE MINING

1.4.1

Risk Factors

Decisions to exploit a particular orebody by block or panel caving methods, the design of caving mines, and the mining operations themselves, involve risks of a number of types. Some of these risks have been referred to in the descriptions of cave mining operations given in Section 1.3 above. Detailed analyses of some of the major risk factors are given in subsequent chapters. Accounts of the risks associated with cave mining have been given recently by Heslop (2000) and Summers (2000a & b). The following is an indicative but not exhaustive list of some of the risk factors requiring consideration at various stages in a cave mining project: •

adequacy of the geological data used in making estimates of the structure, shape, size and grade of the orebody;

•

adequacy of the geotechnical data available about the orebody and country rock masses including major structures, discontinuities, rock properties, in situ stresses and groundwater hydrology. These data are used in making assessments of caveability, cave initiation and propagation, fragmentation, caving performance, excavation stability and dilution;

27

Chapter 1: Introduction

•

caveability assessment usually involving a prediction of the hydraulic radius (area/perimeter) of the undercut at which caving will initiate for a rock mass having given or estimated geotechnical characteristics;

•

cave propagation which is the ability of the cave to continue to propagate once caving has been initiated. Cave propagation depends on a number of factors including the undercut design, the rate of undercutting, the stresses induced on the boundaries and above the cave, the orebody structure and its geotechnical characteristics, and the draw control strategy employed. Because of the capital intensive, non-selective and relatively inflexible nature of caving methods of mining, the inability to initiate or sustain caving is one of the greatest risks faced in cave mining;

•

the degree of fragmentation of the ore occurring as a result of the caving process. This factor influences drawpoint spacing and design, equipment selection and performance, the occurrence of “hangups” and the need for secondary breakage in the drawpoints, the need for underground crushers and the productivity of the cave;

•

caving performance reflects the achievement or otherwise of the planned rate of cave propagation, rate of production, degree of fragmentation, ore grades and recovery;

•

excavation stability refers to the stability over the design life and the need for support or reinforcement of mine excavations including undercut drifts, extraction level excavations, drawbells and items of mine infrastructure. As has been illustrated by the examples given in Section 1.3, excavation stability can depend not only on the geotechnical properties of the rock masses involved and the in situ stress field, but also on factors such as the threedimensional mine layout, the relative timings of certain development and mining activities and the rate of undercutting;

•

major operational hazards including major excavation collapses, mud rushes, rock bursts, air blasts, and water and slurry inflows;

•

environmental risks broadly defined involving issues such as the mine’s influence on surface water and groundwater, the treatment and disposal of mine wastes, influence on flora and fauna habitats, surface subsidence effects, land rights and archaeological issues and other areas of community concern; and

•

risks to profitability arising from factors such as changes to cost structures, industrial relations, variations in metal prices and currency values and local political instability.

28

Chapter 1: Introduction

1.4.2

Introduction to Risk Assessment

Techniques known as risk analysis, risk assessment and risk management are now applied to a wide range of engineering and other undertakings. In the present context, our concern is with the assessment and management of the risks associated with the adoption and operation of a particular mining method. A useful generalised definition of risk assessment is that given by the UK Engineering Council: “Risk assessment is a structured process which identifies both the likelihood and extent of adverse consequences arising from a given activity.” Engineering decisions of the type being considered here are subject to a number of uncertainties, the manifestation of which can result in the failure of a project to meet its objectives in full or in part. These uncertainties can be considered to be of two general types: what we know we don’t know, or parameter uncertainty; and what we don’t know we don’t know, or conceptual uncertainty. Parameter uncertainty is the easier of these two types of uncertainty to account for in engineering procedures. Use of the long established concept of a factor of safety is a commonly used method of attempting to account for parameter uncertainty. Probabilistic methods are also used as an alternative approach to addressing the same issue, particularly in geotechnical engineering (eg Christian et al 1994, Pine 1992). Conceptual uncertainty or uncertainty about how particular sets of conditions will develop and their eventual outcomes, is usually of greater concern and more difficult to address. A risk assessment and management approach seeks to understand the sources of risk associated with a given project or design, to evaluate their consequences and to put in place procedures to manage those risks. Implementing risk assessment and management processes is especially important in the early stages of a potential cave mining project when critical decisions about the adoption of a particular mining method, layout and excavation sequence are being made. In this book, and particularly in Chapter 11, a series of definitions associated with the assessment of risk will be used. Some of these definitions will be introduced here. Hazard - a potential occurrence or condition that could lead to injury, delay, economic loss or damage to the environment. Risk – the product of the probability of occurrence of a hazard and the magnitude of the consequences of the occurrence.

29

Chapter 1: Introduction

Risk analysis – a structured process that identifies both the likelihood and the consequences of the hazards arising from a given activity or facility. Risk evaluation – the appraisal of the significance of a given quantitative (or, when acceptable, qualitative) measure of risk. Risk assessment – comparison of the results of a risk analysis with risk acceptance criteria or other decision parameters. Risk management – the process by which decisions are made to accept known risks or the implementation of actions to reduce unacceptable risks to acceptable levels. The application of these concepts and processes to caving methods of mining will be discussed in Chapter 11. Because the emphasis of this book is on the geomechanics of cave mining, five principal forms of risk will be considered – caveability, fragmentation, caving performance, excavation stability and major operational hazards. Each of these issues has been raised in Section 1.4.1 and will be discussed in detail in subsequent chapters.

1.5

SCOPE AND CONTENTS OF THIS BOOK

This book is intended to provide a digest of the state-of-the-art of block and panel caving from a geomechanics perspective. It reflects the outcomes of the International Caving Study (ICS) Stage I, including the Block Cave Manual (Laubscher 2000) but also contains chapters on several topics that were not part of the ICS research program. Much of the information presented on current caving operations and key caving issues is drawn from papers presented at the international mass mining conference, MassMin 2000, held in Brisbane, Australia, in late 2000 (Chitombo 2000). The subsequent chapters deal with the following topics: Chapter 2 – Rock mass characterisation reviews the needs for and methods of collecting data for use in characterising rock masses for cave mine engineering purposes. The chapter discusses new methods of archiving, correcting, processing and modelling the data developed as part of the ICS. Chapter 3 – Caveability assessment addresses a topic that is vitally important in the study and design of any potential block or panel caving operation. It reviews the available empirical and numerical methods of predicting caveability with an emphasis on the extended Mathews method developed as part of the ICS.

30

Chapter 1: Introduction

Chapter 4 – Fragmentation assessment reviews the factors influencing the fragmentation produced by caving, methods of fragmentation measurement and the available methods of predicting in situ, primary and secondary fragmentation. Chapter 5 – Cave initiation by undercutting discusses the factors influencing undercut design and performance, and the undercutting strategies and undercut designs used in practice. It reports a parametric study of the influence of undercut strategy on the stresses induced in the undercut and extraction levels and the associated support and reinforcement requirements. Chapter 6 – Extraction level design is concerned with the layout and design (including the support and reinforcement) of production and drawpoint drifts, drawpoints, drawbells, pillars and ore handling facilities on extraction or production levels. Chapter 7 –Draw control discusses the importance of draw control in caving mines, the factors influencing the flow of broken ore, the numerical modelling of particle flow, and draw control practice including computer based techniques. Chapter 8 – Geotechnical monitoring considers the nature and purposes of geotechnical monitoring systems and their application to monitoring extraction level and infrastructure performance, surface subsidence and ground movements in caving mines. Chapter 9 – Surface subsidence discusses the mechanisms of caving to surface that may be associated with caving operations and presents methods of analysis of plug subsidence, chimney caving and progressive hangingwall caving. Chapter 10 – Major operational hazards associated with block and panel caving are taken to include major collapses, rock bursts, mud rushes, air blasts and water and slurry inrushes. The causes, effects and means of prevention or amelioration of the effects of these hazards are discussed. Chapter 11 – Risk assessment addresses the increasingly important topic of risk assessment and its application to cave mine engineering. Concepts are introduced, terms defined, the available tools outlined and their application to cave mining described through an account of the tool CaveRisk developed as part of the ICS.

31

CHAPTER 2 ROCK MASS CHARACTERISATION

2.1

DEFINING THE MINING ENVIRONMENT

A

s was noted in Chapter 1, in recent years there has been renewed interest internationally in the mining of massive, often lower grade, orebodies by caving methods. A feature of block and panel caving methods of mining is that, while they have low operating costs, they require high levels of capital investment in infrastructure and development before production can commence. A second feature of these methods is that the mine construction and development required are not readily or economically adaptable to other methods of mining if, for some reason, the chosen mining method proves to be unsuccessful. Accordingly, it is especially important that, when these methods of mining are being considered, the mining environment, especially the geotechnical environment, is understood sufficiently well to permit critical decisions to be made reliably in the pre-feasibility and feasibility study stages of a project. Not to do so invites disaster. As Laubscher (1993) has suggested, the mining environment as it is being identified here must be defined, or re-defined, in a number of circumstances: •

for new mining projects on greenfield sites;

•

for planning the mining of new mining blocks or orebodies in current operations at established mines;

•

where a change from open pit to underground mass mining is being considered; and

•

where difficulties encountered in operations require a review of the current mining method, planning parameters, layout or detailed mine design.

Examples of each of these circumstances have been encountered in cave mining projects and operations in recent years. In each of these circumstances, data of the types being considered here may be required for the following major purposes: 32

Chapter 2: Rock Mass Characterisation

•

mining method selection including caveability studies;

•

detailed design of mining excavations including their sizes, shapes and requirements for support and reinforcement;

•

fragmentation studies which influence issues such as drawpoint spacing and design and equipment selection (including crushers);

•

production or extraction level layout and detailed design including support and reinforcement requirements;

•

mine infrastructure location and design;

•

impacts of mining on the surface including the nature and extent of caving zones, interactions with water courses or storages, impacts on surface installations, and impacts on local communities; and

•

risk assessment, especially for major hazards such as mud rushes, rock bursts, major instabilities and associated air blasts.

The data required for these purposes can be considered as falling into several categories. The general requirements in each of these categories will be outlined in Section 2.2.

2.2

2.2.1

GENERAL DATA REQUIREMENTS Geology

It may be assumed that the regional geology will have been assessed during the exploration stage of a new project or will be well established and understood on continuing projects (eg Howell and Molloy 1960). The local mine geology must be known and understood in some detail for the purposes identified above. Knowledge is required of issues such as: •

orebody shape, size and the distribution of grades;

•

location and nature of the contact of the orebody with the country rock;

•

the nature of the country rocks and of any weathered or transported overburden materials; and

•

structural features such as faults, shear zones, dykes, sills and folding.

33

Chapter 2: Rock Mass Characterisation

Generally, the information available from exploration drilling is incomplete with the result that planning and production engineers may find themselves "mining blind" (Hood et al 1999). Poor detailed knowledge of the orebody geometry in underground metalliferous mines can result in dilution or incomplete ore recovery or both. Developing the ability to "see" through the rock mass in order to gain a more detailed knowledge of the ore grades and boundaries, and of the rock structure and strength, would bring immense benefit. Geophysical techniques using seismic and electromagnetic methods, for example, are considered likely to provide an effective means of supplementing the information available from drilling (Hood et al 1999). It is especially important that both major and minor faults and shear zones intersecting the orebody and the nearby country rock be identified. The classification of faults and shear zones is considered in Section 2.3.1 below. The potentially deleterious effects of faults intersecting, or in close proximity to, mining excavations have long been recognised and dealt with. The nature and magnitudes of these effects may vary with the orientation of the fault, the geomechanical properties of the adjoining rock, the nature of the fault material and fault surfaces (friable or broken material, clay or other filling, slickensiding), the size of the excavation and the presence of water. Some of the observed effects of faults on underground mining excavations include: •

off-setting of the orebody;

•

slip on the fault leading to a re-distribution of stresses around the excavation;

•

fretting or chimneying of friable fault and surrounding material above the back or hangingwall;

•

isolation of large blocks or wedges that become free to slide or fall into the excavation;

•

general sloughing of destressed or unrestrained rock leading to dilution;

•

inability to form a satisfactory anchorage for, or to complete the installation of, reinforcing elements such as rock bolts and cable bolts; and

•

the provision of a conduit for water flows into the excavations.

Sourineni et al (1999) recently carried out a study of fault-related sloughing in open stopes and gave several examples of major fault-induced failure. Heslop (2000) points to several effects of faults in block caving operations while Laubscher (2000) notes particularly the potential for faults to isolate large wedges which may "sit down" on major apices or drawpoints and inhibit uniform draw and cave propagation.

34

Chapter 2: Rock Mass Characterisation

2.2.2

Surface and Groundwater Hydrology

Surface and groundwater management is of little concern in some caving operations but is vitally important in others. It is necessary, therefore, that issues such as the location of surface water courses and storages, rainwater drainage and groundwater hydrology (including the potential for recharge) are evaluated in the feasibility study stage. If in caving operations, the ingress of water into the caving zone can be prevented, the mining excavations will serve to drain the surrounding rock mass with the result that the mine will be dry. However, in other cases, including areas of extremely high rainfall and where there are adjacent water storages and tailings dumps, water control and mud rush problems can be of concern (eg Barber et al 2000, Butcher et al 2000). These issues will be discussed in Chapter 10. 2.2.3

Topography and Environmental Constraints

The topography in the area of the orebody will have a major influence on the locations and costs of surface infrastructure and underground accesses. The local topography will also have an influence on the hydrological issues just discussed, on the local in situ stresses (see Section 2.8) and on the way in which any caving zone eventually propagates to surface (eg Brown and Ferguson 1979). Obviously, the existence of communities and of utilities such as roads, power lines and pipe lines of various types in the zone likely to be affected by the mine must be established and taken into account. There are many examples of the positive impacts of new mining projects on local communities by providing jobs, improved services and custom for local businesses. The social and environmental impacts of mining have become of major concern in recent years. They are noted here for completeness and will be considered no further. A particular issue in some parts of the world, including Australia, is native land rights and the existence of sacred and archaeological sites in the area influenced by mining. Expert studies of all of these issues are usually required to inform the definition of the mining environment of any mass mining project. The Century Zinc Project in North West Queensland, Australia, provides an especially good example of the successful resolution of issues of this type (Williams 1999). 2.2.4

Geotechnical Studies

Most of the key issues referred to in Section 2.1 outlining the uses of the data required to define the mining environment, require geotechnical data for their resolution. Accordingly, the emphasis of the remainder of this chapter will be on the collection and assessment of geotechnical data for rock mass characterisation. As well as the general geological and hydrogeological data referred to above, the geotechnical data required for the purposes being considered here includes:

35

Chapter 2: Rock Mass Characterisation

•

discontinuity survey data obtained through core logging, downhole logging of boreholes or scanline mapping of exposed faces. The data required are the locations, orientations, nature and condition of all discontinuities encountered and, in exposures, their terminations. These data are vitally important in caveability, fragmentation and excavation stability studies;

•

measurements of the physical and mechanical properties of the lithological units making up the orebody and the immediate country rocks. These include - unit weights, - uniaxial compressive and tensile strengths, - shear strengths of discontinuities, - shear strength parameters of intact rocks, - stiffnesses and deformation moduli of discontinuities and intact samples, and - hardness, toughness, abrasivity and drillability indices.

Standard methods of measuring these properties are given by Brown (1981). •

rock mass classification of all lithological units using one of the established methods such as those due to Barton et al (1974), Bieniawski (1976) or Laubscher (1977, 1994). These values are used in caveability studies, empirical methods of stability assessment and in estimating rock mass strengths using methods such as those developed by Laubscher (1977, 1994) and Hoek and Brown (1980, 1997); and

•

measurements or estimates of the regional and mine site in situ stresses. The stresses induced around mining excavations have major influences on excavation stability and, importantly in the current context, on cave propagation (Kendrick 1970, Krstulovic 1979, van As and Jeffrey 2000).

This geotechnical data collection phase is not always carried out adequately in terms of the nature, quantity or quality of the data collected or the time at which it becomes available for use in feasibility and subsequent mine design studies. The remainder of this chapter will concentrate on geotechnical data collection and its use in rock mass characterisation. A general review of the field will be given with emphasis on those topics to which particular contributions have been made as part of the International Caving Study Stage I as reported by Harries (2001).

2.3 2.3.1

CLASSIFICATION AND DESCRIPTION OF DISCONTINUITIES Classification

The term discontinuity refers to any mechanical discontinuity in a rock mass having zero or low tensile strength. It is a collective term for most types of joints, weak bedding planes, weak

36

Chapter 2: Rock Mass Characterisation

schistosity planes, weakness zones and faults. It contrasts with more specific terms such as bedding plane, joint or fault which describe discontinuities formed under particular conditions and mechanisms. Rock mass discontinuities can be classified using a number of criteria. A useful geometric classification used by structural geologists is to describe the discontinuity as a penetrative or non-penetrative structure. A particular structure is said to be penetrative if, on the scale under consideration, that structure is repeated with much the same spacing and orientation pattern, from one sample of the rock mass to the next. Otherwise the structure is non-penetrative; that is, the same structure may occur within different samples in different regions of the rock mass, but its distribution, spacing and orientation are not similar from one sample to the next (Hobbs 1993). Penetrative rock mass structures that can influence the mechanical properties and hence the behaviour of a rock mass include the bedding surfaces of sedimentary rocks, flow foliation of igneous rocks and foliation of regionally deformed metamorphic rocks. Typical non-penetrative discontinuities that influence the rock mass behaviour are joints and faults. It has been argued by Hobbs (1993) that "many engineering geologists have a preoccupation with joint surfaces as potential failure planes, to the exclusion of all other structures. While such a preoccupation with fracture systems is probably safe enough in weakly deformed sedimentary and igneous rocks, it can potentially lead to disaster if a keen appreciation of the penetrative structures in metamorphic rocks is not also actively maintained". Penetrative structures such as slaty cleavage or schistosity could be important in determining mechanical responses such as caveability and fragmentation. For engineering applications the most useful geometric classification of discontinuities is by scale. Discontinuities can be divided into two classes by size (Cruden 1977): •

major discontinuities such as faults, dykes, contacts and related features with a size of the same order of magnitude as that of the site to be characterised. The position in space, physical properties and geometrical characteristics are usually established deterministically for each of these major discontinuities; and

•

minor discontinuities such as joints, minor shears and bedding planes which, for practical purposes, represent an infinite population in the area of design. As a result, their geometrical characteristics and physical properties must be estimated by measurements of a representative sampled (smaller) population.

This division of discontinuities is important as it separates those features that may be represented deterministically from those that must be represented statistically. For minor discontinuities, representative sampling, sample size and the definition of structural domains are important aspects of a rock mass characterisation program. Structural domains are zones 37

Chapter 2: Rock Mass Characterisation

of a rock mass in which the geometrical and physical properties of the discontinuities can be treated as being statistically homogenous. Discontinuities may also be classified according to their origins. Joints (minor fractures) Joints are the most common result of brittle fracture in the Earth’s crust. They are ubiquitous geological structures, occurring in a wide variety of rock types and tectonic environments. They also have a dominant effect on the mechanical and hydrological responses of rock masses to engineering activity. In discussing minor fractures (joints), Price and Cosgrove (1990) comment that one of the topics that bedevils fracture interpretation is nomenclature. Griggs and Handin (1960) classify discontinuities as faults if they exhibit shear displacement and as joints if they are dilatational features which exhibit no shear. In practice, this approach could lead to the misinterpretation of geological data. The shortcoming of the definition is that the scale at which the observation is made is not specified. Geologists who are unable to demonstrate a shear displacement on a fracture may be tempted to classify such a fracture as a ‘joint’ resulting only from extension.

Quartz

Calcite

Figure 2.1: Photomicrograph of quartz and calcite filled discontinuities (Price and Cosgrove 1990) Figure 2.1 shows a photomicrograph of intersecting micro-fractures. The photomicrograph of the rock section under crossed polarised light quite clearly shows two quartz filled fractures (both showing undulose extinction) oriented vertically and a single horizontal calcite filled fracture showing high order birefringence. From this photomicrograph, it can be seen that the calcite-filled vein displaces the quartz filled veins (by approximately 0.025 mm). This small displacement can only be inferred because of the infill of the fractures and the use of 38

Chapter 2: Rock Mass Characterisation

microphotography. The two earlier quartz veins are dilational (ie extensional) features and the later calcite vein is a hybrid extension/shear fracture. It is likely that from a field observation, the shear movement would be missed and all three fractures would be identified as joints using Griggs and Handin’s definitions. On the basis of field experience, Harries (2001) has suggested that, in order to avoid such problems, the term joint should be used to describe any minor fracture which does not exhibit a shear displacement of greater than 0.01 m (10 mm). The term fault is used where a shear displacement of greater than 0.01 m is observed along the discontinuity plane during mapping or core logging. Joints often display spatial and orientational relationships with folds, anticlines, synclines and faults formed during some period of tectonic activity (Price 1966, Price and Cosgrove 1990). An example of the jointing associated with folding is shown in Figure 2.2. Veins, or cemented joints, as illustrated in Figure 2.1, are mineral infillings of joints or fissures. They may be sheet-like or tabular or irregular. They are generally of igneous origin but may also result from sedimentary processes. They are commonly associated with metalliferous orebodies and have been found to have important influences on orebody caveability and fragmentation. They may be weaker or stronger than the wall rock of the joint or fissure and should be taken into account in rock mass characterisation schemes (eg. Laubscher and Jakubec 2000). For example, at the El Teniente mine, Chile, the stronger primary ore contains large numbers of veins or filled joints. These veins are described in order of decreasing strength as (Flores and Karzulovic 2002a): • • • •

quartz filled joints; anhydrite-chalocpyrite filled joints; anhydrite-gypsum filled joints; or joints with soft fillings (eg. sericite, oxides).

Figure 2.2: Jointing associated with an asymmetrical anticline (Priest 1993 after Price 1966)

39

Chapter 2: Rock Mass Characterisation

Faults and shears As discussed above, a fault is defined as a discontinuity dividing portions of rock that have been displaced one past the other in shear. Faulting of rock can occur along a single plane or along many planes within a zone. A fault zone is a closely spaced group of parallel or anastomising faults. A zone in which blocks of rock have been displaced but do not display visible fault structures, is termed a shear zone (Hobbs et al 1976). Although shear displacements on discontinuities can range from micrometres to several hundred kilometres, the term fault is reserved for the more extensive features that show significant displacement (>0.01 m, for example). Faults are usually classified on the basis of shear direction, with three main types of fault being defined: • • •

normal faults in which the hangingwall has moved down with respect to the footwall; reverse (thrust) faults in which the hangingwall has moved up relative to the footwall; and strike-slip (transcurrent) faults in which the movement is predominantly sideways along strike.

It is assumed that a fault is induced when changing tectonic stresses produce a shear stress that exceeds the shear strength developed on a particular plane in the rock mass. The type and characteristics of the fault will be controlled by the shear strength of the rock mass and the in situ stress conditions at the time of fault formation. Minor shears and joints will often form as secondary features of the main fault (Kersten 1990). Their orientation, persistence and thickness, and the nature and strength of the infilling materials, influence the effects of faults on mining operations. For a given fault, these features may not be uniform with depth. For example, at the El Teniente and Andina mines in Chile, several examples exist of faults with thicknesses of, say, 1.5 m on the surface, having thicknesses of only 15 cm at depths of 1000 m (Flores and Karzulovic 2002a). Cleavage or schistosity Cleavage or schistosity is predominantly a planar rock fabric (foliation) produced by preferred alignment of platy minerals (generally phyllosilicates). This alignment is not perfect, but is of a statistical nature. Cleavage imparts a special property to the rock, in that it splits preferentially in a direction parallel to the cleavage planes (Ramsay and Huber 1983). The most common type of cleavage is flow cleavage, which is caused by the recrystallisation and realignment of platy minerals during tectonic deformation. This is often associated with low grade regional metamorphism. Fracture cleavage describes incipient, cemented or welded parallel discontinuities that are independent of any parallel alignment of minerals (foliation). Fracture cleavage is another product of high deviatoric stresses developed during tectonic deformation.

40

Chapter 2: Rock Mass Characterisation

Bedding Bedding is typically a continuous plane that divides sedimentary rocks into beds or strata. Similar structures can also occur in igneous rocks from lava flows or the deposition of pyroclastic material. They are created by changes in such factors as grain size, grain orientation, mineralogy or chemistry during deposition. Bedding does not always create discontinuities; in many cases it forms only a slight change in colour or texture in an otherwise intact rock material (Priest 1993). Bedding planes may exist as open fractures or as closed planes along which the rock may part easily (Gerrard 1988). At depth bedding planes are often closed but the release of stress in the rock immediately adjacent to a newly formed excavation may allow bedding planes to part and discontinuities to form (Beer et al 1983). Although initially horizontal and generally planar, bedding can be tilted, folded and even inverted to a complex range of orientations. Bedding features can be recognised by the fact that they are generally parallel, even when tilted or folded. Sometimes there may be minor stratification planes known as cross-bedding, oblique to the major bedding planes. 2.3.2

Description

In the ‘Suggested methods for the quantitative description of discontinuities in rock masses’ of the International Society of Rock Mechanics (ISRM 1978), ten parameters are identified as being required for the quantitative description of discontinuities and rock masses. These parameters are illustrated schematically in Figure 2.3.

Figure 2.3: Discontinuity parameters (Hudson and Harrison 1997)

41

Chapter 2: Rock Mass Characterisation

Five of the ten parameters (orientation, spacing, persistence, number of sets and block size) may be considered to be ‘geometric’ parameters. They will define the geometry of the rock mass structure, the size and shape of rock blocks formed and the nature of intact rock bridges in the rock mass. The other five parameters (roughness, aperture, filling, wall strength and seepage) may be regarded as ‘strength’ parameters because they influence the discontinuity’s shear strength and stiffness.

2.4

DISCONTINUITY DATA COLLECTION BY DRILLING, CORE LOGGING, DOWN-HOLE SURVEYS, SCANLINE AND CELL MAPPING

2.4.1

Introduction

The geometric and mechanical characterisation of discontinuities is a vitally important precursor to engineering design in rock masses. Ideally, the complete characterisation of a rock mass would include a description of each fracture in the rock mass as well as the determination of its geometric and mechanical properties. Currently, this cannot be achieved for a number of reasons: • • •

the visible parts of discontinuities are usually limited to discontinuity traces only; discontinuities distant from the exposed rock surfaces and drill core cannot be observed; and direct and indirect (eg geophysical) measurements of discontinuities have limited resolution and accuracy.

For these reasons, the discontinuities in a rock mass are usually described as an assemblage, rather than individually. This assemblage is modelled stochastically simply because the discontinuity characteristics vary in space (Dershowitz and Einstein 1988). The methods used to collect rock mass discontinuity data can be divided into four main categories: • • • •

geotechnical core logging and borehole imaging; face exposure (planar) mapping; geophysical or indirect mapping; and aerial and photogrammetric techniques.

The fourth category will not be discussed here as it has little relevance to rock mass characterisation for cave mine engineering other than in the exploration stage. Aerial photography can identify major lineaments and the orientations of large-scale discontinuities. Photogrammetric methods have proved useful for rock mass characterisation for open pit mines (Tsoutrelis et al 1990).

42

Chapter 2: Rock Mass Characterisation

2.4.2

Geotechnical Core Logging

The recovery of core by diamond drilling allows information to be obtained from volumes of a rock mass which cannot be observed directly. It is one of the most important and valuable methods of sub-surface exploration and, in some cases, is likely to provide the only direct sampling or observation of much of the rock mass that is to be mined. A large assortment of drilling rigs, core barrels and drill bits are available to provide drill core at diameters from 20 to 150 mm at varying depths and from rocks of varying strength. Hoek and Brown (1980) and Brady and Brown (1993) discuss the principal types of drill rigs, core barrels and drill bits used in diamond drilling. Most of the core drilled during the feasibility and planning stages of a block caving operation will be used to determine the rock types, ore grades, ore textures and geological structures required to develop a geological model of the orebody. Nevertheless, for engineering purposes, valuable information concerning the rock mass can be obtained by a critical examination of core, providing that the examiner is aware of those geological features that are of significance (Deere 1964). The quality of the core record that is obtained and logged is dependent on a number of factors, including: • • • •

the rock strength and behaviour of discontinuities during drilling; the drilling equipment and core diameter used in the coring process; the competence of the drill rig operators; and how the core is handled and stored.

These four factors have been discussed in detail by Onederra (1999) and earlier in the seminal paper by Rosengren (1970). After drilling, wherever possible, the drill core should be oriented before being logged geotechnically. Drill core orientation and the determination of discontinuity orientation are important because of the impact that discontinuity orientation has on the caveability, fragmentation and support requirements of a rock mass. During a drill run, the core will tend to rotate within the core barrel. Thus, the true orientations of discontinuities within rotated core will remain unknown unless they can be correctly reoriented to their original positions within the borehole. The techniques that can be used to assist with the reorientation of core include: • • • •

down-hole orientation instrumentation; the use of reliable reference planes; the use of cameras and geophysical visualisation techniques; and the use of multiple holes and stereographic techniques.

These techniques will be discussed in turn.

43

Chapter 2: Rock Mass Characterisation

Down-hole orientation instrumentation Down-hole orientation devices assist with orienting sections of the core into their correct positions within the drill hole. If the plunge and trend of the borehole are known, the measured orientation of the discontinuity in the borehole can be used to determine its true orientation. Sullivan et al (1992) compiled a useful summary highlighting the advantages and disadvantages of the main down-hole core orientation techniques used in Australia (see Table 2.1). Table 2.1: Core orientation techniques (after Sullivan et al 1992) Technique

Christensen-

Cost*

High

Complexity

Applicable rock

in Use

strength

Moderate

Low

High

Yes

Yes

Hügel

Advantages

Disadvantages

Can be used in low

Requires

strength rocks and

conventional

vertical holes

drilling and is time consuming

Craelius

Moderate

Moderate

Marginal

Yes

**

Simple in use,

Can be subject

negligible drilling

to damage. Not

delays

accurate for high angle defects

Spear

Negligible

Very low

Marginal

Yes

**

Simple in use,

Reliability of

negligible drilling

result

delays Clay imprint

Negligible

Low

Marginal

Yes

**

Acid etch

Moderate

Moderate

Yes

Yes

Simple in use,

Requires

minimal drilling

interpretation of

delays

imprint

Provides core

Time consuming

orientation and additional borehole survey Rocha

Very high

High

Excellent

No

Can be used in

Expensive and

extremely low

requires

strength rocks

specialist equipment

* Both initial outlay and delay in drilling ** Due to top of core run being disturbed during drilling

44

Chapter 2: Rock Mass Characterisation

Sullivan et al (1992) reported that the most common and reliable of the core orientation methods used in Australia was the clay imprint technique. However, there is a trend for drilling contractors to use the down-hole spear technique. The advantages of the spear technique are the almost negligible costs and delays to the drilling operation. As most exploration drilling contractors are paid primarily by the metre drilled, any delay in drilling is seen to be a waste of money. During the geotechnical drilling campaign at Mont Porphyre, Canada, Coulson et al (1998) and Nickson et al (2000) compared five core orientation methods. They found that the Core Tech Canada diamond drill core orientation system, which uses an acid etching technique, gave reliable orientation results. The clay imprint method worked with some success but at depths greater than 1200 m difficulty was experienced in obtaining an adequate imprint. A third system used scribes three lines on the core in a known orientation. Problems were found in using this technique because of the high rock hardness and the induced rotation of the core tube in the barrel. The other methods considered were a borehole camera and the orientation to known bedding, both of which will be discussed below. Because of the depth of the Mont Porphyre orebody (1000 to 1700 m below surface), operational difficulties were experienced in orientating the drill core. The high cost of drilling at such depths and the small number of holes drilled, meant that the cost and time associated with one of the more advanced core orientation systems was acceptable at Mont Porphyre. However, this is unlikely to be the case in other operations which have a greater number of holes, shallower drilling depths and direct underground access to the orebody from exploration drifts, for example, enabling reliable discontinuity orientation data to be obtained. Utilising reference planes of known orientation If the rock mass contains a planar fabric that is consistent throughout the site (eg bedding planes or a regionally consistent cleavage), and if the orientation of the fabric is known, the true orientation of the discontinuity can be established. As long as the reference plane is consistent and the borehole direction (trend and plunge) is known, the angle measured between the reference plane and the discontinuity can be used to determine the dip and dip direction of the discontinuity. The method described by Priest (1985) makes use of the stereographic projection technique. Ideally, another orientation method should be used to check the results obtained using this technique. If there is a major deflection of the reference fabric, perhaps due to faulting or folding, then gross errors can be produced in the estimated discontinuity orientation. Utilising instruments such as cameras or geophysical visualisation techniques Downhole cameras or geophysical techniques that can measure discontinuity orientation can be used to examine the sides of a borehole. Discontinuities intersecting the borehole can be analysed and their orientations determined. These discontinuities can then be related to the

45

Chapter 2: Rock Mass Characterisation

corresponding structures found in the core at the same depths. Although this is a slow process for determining discontinuity orientation, it may be of use where no other orientation methods are available or where processes such as core washing and core blockage have damaged the core to such an extent that traditional orientation methods cannot be used. Using stereographic techniques and multiple holes with different orientations The true orientations of sets of discontinuities can be determined using the orientations of two or more non-parallel boreholes. Using the angle between the core axis and the discontinuity normal, vital information on the discontinuity orientation can be obtained. Rotation of the core during drilling and handling may mean that the true orientation of the discontinuity cannot be determined using this measurement alone. The angle represents a locus around the drilling direction of possible discontinuity orientations. However, using a number of measurements of this angle for the same discontinuity set, made at different drilling directions, the orientation of the discontinuity set can be calculated. Priest (1985) provides a description and an example of the use of this technique to determine discontinuity orientation. Reorienting the core and establishing a reference line are the next steps required in preparation for core logging. Using the core orientation information, the core is rotated into position with the reference line placed on top. Discontinuities or broken pieces of core are put into their correct positions so that the reference line can be extended over discontinuities and drilling-induced fractures. Problems can occur if a large section of broken core or a zone of washed away fault gouge means that the core pieces cannot be fitted together. Geotechnical core logging is intended primarily to provide information on the rock mass discontinuities. There are a number of other basic steps or procedures that should be carried out prior to the detailed geotechnical logging of core, namely: •

the core should be examined to determine the structural boundaries and the geological features to be measured. Markers indicating the depths of the geological horizons and start and end of drill runs should be checked (Brown 1981);

•

core recovery should be established. This parameter may be determined for individual core runs or rock types or for entire boreholes. Recovery in a rock mass of poor quality will be strongly dependent on the drilling equipment used and the skill of the drilling crew; and

•

an assessment should be made of the degree of fragmentation of the drill core. This assessment can be made quantitatively using a number of indices including the Total Core Recovery (TCR), Solid Core Recovery (SCR), Rock Quality Designation (RQD), Fracture Index (FI) and Core Loss (CL) (Windsor and Thompson 1997). The most commonly used measures are the RQD and the Fracture Index (or Fracture Frequency) which find widespread use in rock mass classification schemes and in engineering design.

46

Chapter 2: Rock Mass Characterisation

When determining discontinuity frequency indices care must taken to distinguish between natural discontinuities and fractures caused by the drilling and handling processes. A trained engineering geologist or geotechnical engineer can usually identify natural discontinuities because of the presence of mineral coatings, smoothness, staining and/or weathering of the surrounding rock.

Fully circular

Centre line

Zero

Tip to tip

Fully circular

Centre line

Three methods may be used for determining the lengths of core pieces for RQD calculations tip to tip, centre-line and fully circular (see Figure 2.4). Tip to tip measurements involve double counting at each end of a core piece, while fully circular measurement ignores core pieces that happen to have been drilled with a small subtended angle to one discontinuity in otherwise massive rock. Consequently the centre-line method is recommended (Brown 1981). A discussion of the calculation of RQD is given in Section 2.7.2 below.

Figure 2.4: Different interpretations of the length of core pieces (after Brown 1981) Following the definition of geotechnical domains and the calculation of core recovery and discontinuity intensity measures, the actual logging of the core can proceed. The amount of information concerning rock mass discontinuities that can be obtained from logging drill core will depend on whether or not the core has been oriented. The ten specific parameters described under the ISRM suggested methods (Brown 1981) are listed in Table 2.2. If the core is not oriented, only three of the ten parameters can be recorded reliably. The more disturbance the core has undergone during drilling and handling then generally the less the information that can be obtained on the rock mass discontinuities. For example, care needs to 47

Chapter 2: Rock Mass Characterisation

be taken when assessing discontinuity filling. Discontinuity filling and mineralisation may be washed away by drilling fluids. In some situations drilling mud may be deposited into open discontinuities. Table 2.2: Discontinuity parameters measurable from core record (Harries 2001) Discontinuity Parameter (ISRM 1978) 1. Orientation 2. Spacing 3. Persistence 4. Roughness 5. Wall Strength 6. Aperture 7. Filling 8. Seepage 9. Number of Sets 10. Block Size

Oriented Core

Unoriented Core

Yes Yes No Yes Yes No Yes No Yes No

No (possible) No (possible) No Yes Yes No Yes No No (possible) No

Even when the core has been successfully oriented, the errors associated with sampling in one direction only always need to be considered. A ‘blind zone’ will be present even when a perfect drilling operation is achieved. Only by the use of multiple drill holes at different orientations can this sampling issue be resolved. The main drawback of characterising discontinuities from drill core is the lack of information available on persistence. Because of the small volume of the rock mass contained in the core, it is unlikely that many observations on discontinuity termination will be made or much information obtained on discontinuity shape or size. Optical imaging of the drill hole walls is also used in rock mass characterisation. A variety of techniques are available including photoelectric transformers, conventional cameras operated remotely from a wireline, television cameras operated remotely using coaxial cable and digital borehole scanners. Discontinuity detection applications are sometimes limited because the imaging techniques used require clear borehole fluids, and because they commonly image the borehole at oblique angles using local illumination sources that produce shadows which interfere with discontinuity detection and interpretation (Paillet et al 1991). Furthermore, some data may not be in digital form initially, and so not amenable to sophisticated processing. The digital borehole scanner (DBS) is a logging tool which provides optical, true-colour images of a borehole wall. The Borehole Image Processing (BIP) system was developed by the Raax Company of Kitaku, Sapporo, Japan (Kamewada et al 1990). The digitally recorded data yield high resolution images, enabling detailed measurement of discontinuities, identification of 48

Chapter 2: Rock Mass Characterisation

discontinuity mineralisation, and observation of other microscale properties not generally available with other instruments. However, the sampling issues (orientation blind zone and lack of persistence information) associated with logging oriented geotechnical drill core also apply to borehole imaging techniques. 2.4.3

Exposure Mapping Methods

Mapping of natural (outcrop) or artificial (excavated) rock mass exposures, can enable discontinuity parameters to be characterised in greater detail than is possible from drill core logging. In particular, important information about how discontinuities terminate, the sizes of discontinuities and reliable measurements of discontinuity orientation can be obtained from exposure mapping. Mapping techniques can be divided into three main classes: • • •

spot mapping; lineal mapping; and areal mapping.

The technique used in a given rock mass characterisation campaign depends on the degree of detail and hence the sampling effort required. All techniques will suffer from some degree of sampling bias (see Section 2.5.2), because discontinuities having three dimensional parameters (size, shape and orientation) are being sampled in a two dimensional section (plane). The mapping technique used will determine which analysis techniques can be used to correct for these sampling biases. Spot mapping Spot mapping is a technique in which the observer selectively samples only those individual discontinuities that are considered important. An example of the application of this technique is shown in Figure 2.5. The actual traces observed in the exposure are shown on the left of the figure. The discontinuities depicted on the right of the figure are those that have been mapped because the mapping personnel have considered them important. The value of the results obtained using this technique depends on the judgement of the observer. By recording the characteristics of only those discontinuities thought to be "important", the volume of data, and thus the mapping effort required, are greatly reduced. However, because the user biases the data, the repeatability is poor. Individual users may consider different discontinuities to be important. This may be the case particularly when those performing the mapping are collecting the data for different engineering purposes (eg caveability analysis, fragmentation analysis, cable bolt design). The use of this method is advised only for preliminary ‘reconnaissance’ style data gathering exercises where there is a need to gather some initial orientation data quickly. This can provide useful input into planning more comprehensive and objective mapping exercises (eg by scanline or window mapping).

49

Chapter 2: Rock Mass Characterisation

Original exposure plane

Spot mapping

Figure 2.5: Spot mapping technique (Harries 2001)

Lineal (scanline) mapping Lineal mapping involves measuring or recording the geometric and mechanical characteristics of all discontinuities that intersect a given sampling line. Examples of lineal or scanline mapping techniques are given by Piteau (1970), Priest and Hudson (1981), Villaescusa (1991), Priest (1993), Windsor and Thompson (1997) and Harries (2001). When carried out correctly lineal mapping can provide significant amounts of data from a sample of the rock mass in a structured and objective way. This technique is illustrated schematically in Figure 2.6. The left of the figure shows the original rock mass discontinuity traces and the right of the figure shows those discontinuities selected during mapping. All discontinuities that intersect either of the horizontal or vertical sampling scanlines and are larger than a designated "cut off" limit are mapped and characterised systematically.

Original exposure plane

Line mapping

Figure 2.6: Scanline mapping technique (Harries 2001)

50

Chapter 2: Rock Mass Characterisation

A typical scanline mapping sheet is shown in Figure 2.7. The header information for the scanline survey includes the coordinates of the mapping location, scanline orientation and elevation, wall information, date and personnel involved in the scanline mapping. The scanline (tape measure) is set up on an appropriate clean wall. This may require the wall to be cleaned. If the outcrop is too weathered or the exposure too blast damaged to enable the discontinuities to be mapped, another location should be selected. After selection of an appropriate site, starting at one end of the scanline, every discontinuity that intersects the scanline tape is measured. The discontinuity parameters measured will depend on the scheme adopted. Using the Villaescusa (1991) scanline mapping format, the following information is collected for each discontinuity intersected: • • • • • • • •

distance of intersection along the tape; number of endpoints observed in the plane (0, 1 or 2); discontinuity type (joint, fault, vein, bedding, shear); orientation (dip and dip direction of the discontinuity); roughness (smooth, rough, slickensided); planarity (planar, wavy, irregular); trace length (length of discontinuity seen in the sample plane); and termination types (intact rock, another joint or hidden)

(>20°)

Figure 2.7: Scanline mapping sheet (adapted from Villaescusa 1991) Simple notation and abbreviations are used to optimise the detail of the observations and increase the speed of mapping. Discontinuity parameters are assessed using simple tools and observations. Discontinuity orientation is usually measured using the Brunton or Clar compass.

51

Chapter 2: Rock Mass Characterisation

The trace length is measured using a tape measure and the other parameters can be assessed visually. The ‘remarks’ column can be used for any other general observations or explanation such as the nature of discontinuity mineralisation or alteration. The length of the scanline is normally extended until a prerequisite number of observations are obtained. What is a prerequisite number? Numbers varying from 40 per discontinuity set and from 150 to 350 total discontinuities per scanline have been suggested. Priest (1993) suggests that between 150 and 350 discontinuities should be recorded. The lower number would be sufficient for a ‘simple’ rock mass containing just three discontinuity sets and the higher number for a more complex rock mass containing up to six discontinuity sets. Savely (1972) found that at least 60 observations were required to stereologically define a discontinuity set along a given sample line. Villaescusa (1991) suggests that at least 40 discontinuity observations are required per set to provide a sound statistical database of the discontinuity set characteristics. Areal (window) mapping Areal mapping involves collecting all data from within a specified area of a rock face, often referred to as a ‘window’. Pahl (1981) gives a discussion of the window mapping technique. The preliminaries and measurement techniques for window mapping are the same as those for scanline mapping, except that all the discontinuities that are above a given "cut off" size are recorded (see Figure 2.8).

Original exposure plane

Window mapping

Figure 2.8: Window mapping technique (Harries 2001) Although this method is more time consuming than scanline mapping, the amount of data collected is even greater and in some cases the amount of geometric bias may be reduced.

52

Chapter 2: Rock Mass Characterisation

Geophysical techniques Geophysical techniques can be used to gain a measure of the geological sequence and structure of the rock mass. These techniques measure various geophysical properties of the rock mass that can be related to its lithological and geotechnical properties. Most of these geophysical methods can measure discontinuities only indirectly. Typically, the reduced data from each method (eg seismic travel times) must be inverted to yield estimates of local rock properties (eg seismic velocities). Normally these rock properties are not discontinuity properties (eg discontinuity density). Rather, the discontinuity properties must be deduced indirectly from the rock properties. This deduction requires strong idealisations of discontinuity geometry. These discontinuity properties are not always the properties of direct interest in many engineering applications (eg discontinuity intensity or in situ block size). The required properties must be interpreted from the deduced fracture properties. This interpretation requires higher levels of subjectivity than the first (inversion) or the second (deduction) steps (National Research Council 1996). Geophysical methods naturally divide themselves into three distinct scales: • • •

large scales associated with surface soundings; intermediate scales associated with surface to borehole and borehole to borehole soundings; and small scales associated with measurements made on rock immediately adjacent to a borehole or tunnel.

Table 2.3: Geophysical discontinuity detection methods (after National Research Council 1996) Method

Length scale (resolution)

Remarks

Elastic methods: seismic band (10-100 Hz)

100 to 5000m

zero shear modulus in fracture fluid is critical

Elastic methods: sonic band (2-20 Hz)

0.1 to 10m

zero shear modulus in fracture fluid is critical

Elastic methods: ultrasonic band (200-2000 kHz) Electrical methods

0.1 to 10m

fracture aperture is critical

10 to 300m

contrasting resistivity of fracture fluid is critical

Electromagnetic methods Radar methods

10 to 300m 3 to 200m

contrasting resistivity of fracture fluid is critical contrasting resistivity of fracture fluid is critical

Conventional well logs

0.1 to 10m

near borehole environment

Large scale geophysical measurements such as those used in the petroleum industry can be used to identify major discontinuities. Major discontinuities such as large fracture zones and faults 53

Chapter 2: Rock Mass Characterisation

can be characterised by seismic, electrical and electromagnetic methods. Minor discontinuities are extremely difficult to detect using these large scale geophysical methods. Intermediate scale methods involving surface to borehole and borehole to borehole soundings have many advantages over the larger scale surface geophysical methods. In many cases, overburden acts as a filter because of its attenuation properties and high contrast, requiring the use of complex correction procedures to obtain useful information. Boreholes and tunnels provide access to measurement points below the surface, allowing many of the problems arising from overburden to be avoided. Compared to surface surveys, borehole measurements sometimes require complex (and compact) sensors. Remote sensing is done with sources and receivers placed in the same borehole, or one in the borehole and the other at the ground surface (National Research Council 1996). Geophysical systems consisting of a downhole probe (or tool) attached to a multi-conductor electric cable often referred to as a ‘wireline’, can also be used to characterise rock masses. The wireline is attached at the surface to a winching assembly that controls the lowering or raising of the probe. Generally, several types of geophysical devices are combined to form one downhole logging tool. The main types of geophysical wireline logging tests include seismic velocity, seisveiwer, electrical resistivity, gamma-gamma and self potential. The applicability of these techniques in assessing discontinuity parameters is outlined in Table 2.4. None of the measurement techniques directly measure discontinuity parameters but some have a correlation between the test result and the discontinuity parameter. Table 2.4: Wireline logging techniques (after Windsor and Thompson 1997) Type of test Seismic

Discontinuity Parameter Orientation

Spacing

Type

Separation

Filling

no correlation

some

no correlation

no correlation

no correlation

no correlation

direct

some

correlation

correlation

Velocity

correlation

Seisviewer or

direct

direct

acoustic

correlation

correlation

scanner Electrical

some

some

resistivity

correlation

correlation

Gamma-

no correlation

some

gamma Self-potential

no correlation

no correlation

no correlation

no correlation

correlation no correlation

some correlation some correlation

no correlation

no correlation

no correlation

no correlation

In all cases, the emphasis on using the above techniques is to determine the properties of the rock, the in situ locations, orientations and nature of the discontinuities and the hydrological conditions that exist within the rock mass. To ensure the best results, these techniques should

54

Chapter 2: Rock Mass Characterisation

be compared with descriptions of sections of the core, and if necessary compared directly with the core. Rock material properties that may be needed in the geophysical data reduction can then be obtained. This allows the geophysical testing to be calibrated to the observed rock mass conditions. Recently, the acoustic scanner has been shown to be an excellent tool for providing oriented images of borehole walls at very fine resolution (Hatherly and Medhurst 2000). Hatherly (2001) suggests that the geophysical tools of relevance in the detection and measurement of defects are acoustic scanners (televiewers), optical scanners (for dry holes), dip meters and full waveform sonic logs. Those methods which rely on the detection of waterfilled discontinuities by electrical and electromagnetic methods may not always be suitable for hard rock mining applications.

2.5 2.5.1

ANALYSIS AND PRESENTATION OF DISCONTINUITY DATA Introduction

The objective of rock mass discontinuity analysis is to estimate the characteristics of the discontinuities and to determine how they vary over the mine site. The estimate of rock mass discontinuity properties or rock mass model, is merely one representation of the results from the collected sample data. Unfortunately, the true characteristics of the rock mass discontinuities can never be known exactly. This would require all of the discontinuities in the rock mass to be measured accurately and their engineering properties established by testing. Major discontinuities such as faults, dykes, geological contacts and unconformities may be analysed deterministically in cave mine design. For all practical design purposes, minor discontinuities represent an almost infinite population. Their characteristics will be assessed by analysis of a much smaller sample population. The objective of this statistical sampling is to infer the characteristics of a large population without including all its members in the sample. It is the statistical treatment of these minor discontinuities that will be discussed here. In order to establish the best estimate of discontinuity properties, the measurement of discontinuity characteristics should be unbiased and objective. A number of biases introduced during sampling should be accounted for either before or during the analysis process. The method of data collection is the most important factor influencing the objectivity of the discontinuity analysis; some methods are inherently more subjective than others. This subjectivity in the data collection is reflected in possible errors in the analysis. It is important to realise that it is much harder to account for error due to subjectivity than for systematic sampling errors.

55

Chapter 2: Rock Mass Characterisation

2.5.2

Error and Uncertainty in Discontinuity Analysis

Einstein and Baecher (1982) have defined three main sources of uncertainty and error in engineering geology and rock mechanics: • • •

the innate, spatial variability of geological formations, where wrongly made interpretations of the geological setting may be of significant consequence; errors introduced in measuring and estimating engineering properties, often related to sampling and measurement techniques; and inaccuracies introduced in modelling physical behaviour, including the use of incorrect or inapplicable types of calculations or models.

Uncertainty and error will be associated with any rock mass characterisation study. While uncertainty can never be removed completely, it should be reduced to a minimum. An acceptable level of uncertainty can be quite difficult to establish. In an engineering study, what has been left out of the analysis cannot always be known. According to Einstein and Baecher (1982) most of the major failures of constructed facilities have been attributed to omissions. Furthermore, real rock masses will have properties and variability that can never be accounted for fully in characterisation, rock mass classification or design analyses. Uncertainties caused by inherent geological spatial variability The geological subsurface is spatially variable in that it is composed of different materials which are stratified, truncated, and in other ways separated into more or less discrete zones. It is impossible to account for all zonation of the rock mass because much of it will be unknown, especially in cave mine engineering where a majority of the orebody will have been characterised from drill core only. This limited sampling will hinder the development of a highly detailed geological model. The other spatial variability occurs within an apparently homogenous body because material properties may vary from point to point. While this variability can be precisely characterised with sufficient observations, the numbers of observations available are usually limited. Thus, uncertainty will remain about the material properties or classifications at points that have not been observed or sampled. Error arising from variability in the geology can never be avoided. It can be reduced by the use of experienced geologists having extensive knowledge of the region concerned and by directing specific investigations towards possible key geologic structures (Palmstrom 1995). Measurement errors and sample biases The main measurement error that occurs in discontinuity characterisation is in the measurement of discontinuity orientation. West (1979) carried out a study of measurement errors in discontinuity mapping by examining the reproducibility of frequency and orientation measurements made during scanline mapping at a quarry in the Lower Chalk in Oxfordshire, 56

Chapter 2: Rock Mass Characterisation

England. He concluded that the orientation of a well-defined joint could be determined to within about ±6° for dip direction and ±5° for dip angle. Subsequently, Ewan and West (1981) examined the reproducibility of joint orientation measurements using six different observers performing 10 m long scanline surveys. They found that the orientation of particular joints recorded can have a mean maximum variation of ±10° for dip direction and ±5° for dip angle without there being any real difference in the joint orientation. However, joint orientation diagrams created for the collected discontinuity data identified the same major joint sets irrespective of observer. In this case, many of the measurements were taken in a full-face machine bored tunnel having smooth walls. It is difficult to obtain measurements of discontinuities from smooth walls because there may not be sufficient discontinuities exposed to enable accurate orientation measurements to be made directly. In these cases a notebook or other planar device is used to project the discontinuity plane from which a measurement can be made. There are four main types of sampling bias in discontinuity measurement: •

Orientation bias – the frequency of discontinuities intersecting a particular window, scanline or drill core depends on the orientation of the sampling geometry relative to the orientation of the discontinuity set. If a discontinuity set is oriented parallel to a window, then few discontinuities in this set will intersect the window.

•

Size bias – the larger the scale of a discontinuity the more likely it is to be sampled by a given drill core, scanline or mapping window.

•

Truncation bias – a truncation or size cut-off is usually used in scanline or window mapping. For example, fractures that are less than 50 mm in length may be ignored. Although using such a small cut-off will usually have little effect on the overall discontinuity statistics, if a comprehensive, rigorous analysis is undertaken with the aim of fully describing the distribution of discontinuity sizes then the truncation size cut-off must be taken into account.

•

Censoring bias – this bias is associated with the artificial boundaries that are imposed when performing rock mass characterisation. Typically in underground mines the most limiting boundary is the height of the drives in which mapping takes place. The restriction in height of the mapping window limits the trace lengths that can be observed. Censored trace lengths provide a lower bound estimate of the true trace lengths.

For an unbiased structural analysis, the measurement process needs to be objective. Biases can be reduced by the appropriate selection of rock mass characterisation methodologies and sites. For example, if three mutually orthogonal scanlines are used, the bias due to orientation will be greatly reduced. However, it is not possible to account for all biases, which must then be removed using analytical or numerical methods. 57

Chapter 2: Rock Mass Characterisation

Modelling errors Models used in assessing discontinuity parameters and modelling rock mass discontinuity geometry are based mainly on observations of discontinuities and a number of assumptions made about the geometry and characteristics of the discontinuities. As they are simplifications of reality, modelling errors are introduced. An example of this kind of error is the use of a rock mass model that models all discontinuities as orthogonal planes of infinite persistence (eg Snow 1965). However, it must be recognised that a degree of simplification is required in the development of any model. 2.5.3

Discontinuity Orientation Analysis

The main aim of an orientation analysis is to establish a statistical model of the orientational arrangement of the discontinuities contained within the rock mass. The underlying premise is that geological processes have generated one or more sets (or clusters) of nearly parallel discontinuities in the rock mass. From the discontinuity orientation data collected, a statistical model that represents the discontinuity orientation characteristics of the rock mass can be constructed. The discontinuity orientation characteristics that are of most interest in rock engineering include: • • • •

the number of discontinuity sets in the rock mass; the mean orientations (dip and dip direction) of these discontinuity sets; the spread or dispersion of orientations around a given set’s mean orientation; and the amount of data that lies outside defined discontinuity set limits.

The first step in the analysis is to determine the number of discontinuity sets in the rock mass and to define the limits of those sets. The main technique used in the identification of discontinuity sets involves presentation of the data in a graphical format. Prior to the widespread use of computers for the plotting and contouring of orientation data, orientation analyses were conducted manually. This manual technique is described by Hoek and Bray (1974) and Priest (1985). A number of computer programs have become available to assist in the plotting and analysis of orientation data, including DIPS (Rocscience Inc. 1999), SAFEX (Windsor and Thompson 1990) and CANDO (Priest 1993). Whether a manual technique or a computer program is used, the best way of representing orientational data graphically is with respect to the surface of a reference sphere using the stereographic projection. Discontinuity orientation data are usually represented by unit vectors normal to the discontinuity planar surface. The discontinuity unit normals are recorded unambiguously by polar coordinates using two angles, the trend, α, and the plunge, β. In the development of rock mechanics and rock engineering, there has been an almost total adoption of equal-angle lower hemisphere projection (Hudson and Cosgrove 1997). An example of an equal-angle lower hemisphere projection is shown in Figure 2.9.

58

Chapter 2: Rock Mass Characterisation

Figure 2.9: Example of a stereographic projection of discontinuity unit normals (from Rocscience Inc. 1999) The discontinuity normals (or poles) can be seen to cluster into three distinct groups which represent discontinuity sets. These independent discontinuity sets need to be identified and analysed separately. Design set boundaries for each of the discontinuity sets need to be identified. A technique that can aid in the definition of discontinuity sets and their boundaries involves contouring the data. The data shown in Figure 2.9 have been contoured using the widely used DIPS program (Rocscience Inc. 1999) with the results shown in Figure 2.10. The same discontinuity sets can be identified in the contoured data. A representative or mean orientation of the discontinuity set can then be calculated. The Fisher (1953) distribution is usually used to characterise the distribution of discontinuity orientations about some ‘true’ mean. It is relatively easy to implement and provides a measure of dispersion about the mean discontinuity orientation, called Fisher’s constant, K. It is, however, a symmetric distribution and therefore provides only an approximation for asymmetric data. Watson (1966) and Einstein and Baecher (1983) provide a number of asymmetric models, such as the Bivariate Fisher, which can provide better fits for asymmetric orientation data. A detailed discussion of the Fisher distribution is given in Section 2.6.3. A study by Dershowitz and Einstein (1988) conducted on several distributions using data from a number of sources concluded that none of the currently used distributions were statistically acceptable in all cases. The Fisher, Bivariate Fisher and Bingham distributions provided equal numbers of good fits. The added complexity of asymmetric orientation distributions does not improve the fit of real data to modelled orientation distributions.

59

Chapter 2: Rock Mass Characterisation

Set 2

Set 3

Set 1

Figure 2.10: Contoured stereographic projection (from Rocscience Inc. 1999)

2.5.4

Discontinuity Frequency/Spacing Analysis

The discontinuity frequency is a fundamental measure of the degree of fracturing of a rock mass. Discontinuity frequency can be expressed as the number of discontinuities observed or predicted within a unit volume (volumetric discontinuity density), a unit area (areal discontinuity density) or a unit length (linear fracture frequency). Discontinuity spacing is a measure that is linked to the discontinuity frequency. At its simplest, the discontinuity spacing is the distance between one discontinuity and another, or the reciprocal of the linear fracture frequency. The volumetric discontinuity density, λv, is the most fundamental of the three measures of discontinuity intensity (Priest 1993). It is based on the assumption that discontinuities can be represented by the occurrence of a point located at the centroid of the discontinuity. The volumetric frequency, λv, is the average number of points per unit volume of the rock mass (m-3). The measure can be applied to all the discontinuities contained in the rock mass, λv, or individually for the density of each discontinuity set, λvn, where n represents the discontinuity set number. Although λv is an attractive measure of discontinuity intensity, its direct measurement would require the rock mass to be dissected in a non-destructive manner. This is currently impractical. Accordingly, the volumetric frequency must be estimated from areal or lineal density measurements, following methods described by Baecher et al (1977), Warburton (1980) and Villaescusa (1991).

60

Chapter 2: Rock Mass Characterisation

Discontinuity areal frequency, λa, is the average number of points that occur in a unit area of a section through the rock mass. This parameter can be measured from a sampling plane using a wall mapping technique. The problem of using an areal intensity measure is the need to consider the possible orientation bias of the sample plane. A simple geometric correction introduced by Terzaghi (1965) can be used to relate the true discontinuity frequency to the apparent discontinuity frequency:

λ as = λ a sin γ

(2.1)

where λas is the apparent areal discontinuity frequency and γ is the angle between the mean discontinuity plane orientation and the vector normal to the sample plane. There are also sample biases associated with censoring and truncation effects imposed by exposures of limited extent and issues associated with the shapes of discontinuities. As a result of these complexities, areal discontinuity frequency is not a measure of discontinuity intensity that finds much practical use. The linear fracture frequency is the simplest and most commonly used measure of discontinuity frequency. It is used in the MRMR rock mass classification system (Laubscher 1990), in estimates of in situ block size (Palmstrom 1996) and is recorded in drill core logging. Linear fracture frequency is also often used to estimate the RQD parameter (needed in most classification systems, see Section 2.7) in a correlation proposed by Priest and Hudson (1976). The widespread use of linear fracture frequency as a measure of discontinuity intensity owes much to the use of scanlines and drill core as the major discontinuity characterisation techniques. The linear fracture frequency, λl, will be dependent on the orientation of the sampling line unless the discontinuity network is isotropic, which is unlikely. The linear fracture frequency of a particular discontinuity set n, λln, represents the linear fracture frequency of the set perpendicular to the discontinuity plane (in the direction of the mean unit normal). The apparent linear fracture frequency for a particular discontinuity set) λlsn, will be that sampled by a scanline or drill hole oriented in a particular direction (see Figure 2.11). The true and apparent linear discontinuity frequency are related by the Terzaghi (1965) correction given by Equation 2.1 with γ being the angle between the mean discontinuity plane orientation and sample line direction. Using this result and by combining the results of the number of discontinuity sets contained in the rock mass, the estimated rock mass fracture frequency (λ) for different sampling directions can be estimated. Discontinuity fracture frequency data for a rock mass, in particular the fracture frequency extrema (maximum and minimum fracture frequencies), have been investigated by Hudson and Priest (1983) who produced some useful results for obtaining loci of discontinuity frequency.

61

Chapter 2: Rock Mass Characterisation

Apparent discontinuity set fracture frequency (λlsn)

True discontinuity set fracture frequency (λln)

Figure 2.11: True and apparent discontinuity set linear frequency Discontinuity spacing is a widely used measure of discontinuity frequency. It is used in all the rock mass classification schemes and a number of the techniques used to predict in situ block sizes outlined in Chapter 4. A number of different measures are used to define discontinuity spacing. The ‘Normal Set Spacing’ is the perpendicular distance between sub parallel discontinuities from the same set (the reciprocal of λln, see Figure 2.11). The ‘Apparent Set Spacing’ is the spacing between a pair of immediately adjacent discontinuities from a given discontinuity set, measured along a line of any specified location and orientation. The ‘Rock Mass Spacing’ is the spacing between a pair of immediately adjacent discontinuities (regardless of what discontinuity set they belong to) measured along a sampling line. Although the mean spacing of a discontinuity set or of the whole rock mass does provide a useful measure of discontinuity intensity, a greater understanding of rock discontinuity properties can be gained from investigating the full distribution of discontinuity spacing. If a sufficiently large number of individual spacing values are obtained (Hudson and Harrison (1997) suggest more than 200 individual measurements), they can be plotted in histogram form to gain an understanding of the shape of the distribution. On the basis of field measurements, Priest and Hudson (1976) concluded that the distribution of total discontinuity spacings for a variety of sedimentary rock types could be modelled by the negative exponential probability density distribution. This finding has been supported by other investigators (eg Call et al 1976, Einstein et al 1980, Baecher 1983) who worked on a variety of igneous, sedimentary and metamorphic rocks. If the occurrence of a discontinuity along a scanline or drill core is entirely random, then the location of one discontinuity has no influence upon the location of any other. In this case the discontinuity intersections are said to obey a one-dimensional Poisson process. When this 62

Chapter 2: Rock Mass Characterisation

occurs the probability density distribution of total discontinuity spacings is negative exponential with a mean spacing of 1/λl. The fact that a large number of discontinuity set spacings examined follow the negative exponential distribution suggests, but does not prove, that in a statistical sense discontinuity occurrences are random. Hudson and Harrison (1997) emphasise the fact that discontinuities are not random events, even though the Poisson process of random events can be expected to apply to field data. They suggest that spacing values converge when successive spacing distributions of any type are superimposed on the sampling line. The negative exponential distribution is expected as a result of a suite of superimposed geological events, each of which produces fracturing of a given distribution. Roleau and Gale (1985) similarly analysed an extensive discontinuity spacing data base from drill core from a granitoid rock mass. They evaluated the goodness-of-fit of three statistical models, the negative exponential, the log-normal and the Weibull distribution. The results quite clearly showed that the negative exponential distribution did not fit their data but that a log-normal distribution fitted the data very well. Other researchers (eg Bridges 1975, Barton 1977) have also fitted logarithmic spacing distributions to their observations. In reviewing the work of Rouleau and Gale (1985), Mohajerani (1998) noted how the data points which did not fit the negative exponential distribution had mean spacings values of about one metre or larger. He suggested that the distribution of spacing values may be assumed to be either negative exponential or log-normal, depending on the rock type and spacing range (between the maximum and minimum values). This tends to agree with the the superposing theory of Hudson and Harrison (1997); if there are enough geological events to create a number of discontinuity sets and a small total spacing, then the spacing distribution will follow a negative exponential distribution. Where only a few geological events have caused fracturing, or existing discontinuity sets have become healed, a larger total spacing and log-normal distribution of discontinuity spacings may result. Whether a discontinuity spacing distribution follows a negative exponential distribution or a log-normal distribution also depends on the sampling regime adopted. If a truncation level is adopted (ie discontinuities below a certain size are disregarded) then it is likely that small discontinuity spacings will be lost. As a result, a negative exponential distribution would appear as a log-normal distribution. Inaccuracy associated with a calculated mean discontinuity spacing occurs where the estimated value is consistently in error. One particular inaccuracy is that caused by small sampling scanlines. If the length of the scanline is short compared to the mean spacing then a biased result will be obtained. This type of bias can be produced in areas where only short scanlines can be established. In particular, this is a problem in vertical scanlines that are carried out in drives and tunnels. The effect of short scanlines on mean discontinuity spacing calculations is discussed by Sen and Kazi (1984). They produce graphs that illustrate the effect of scanline length on the calculation of mean discontinuity spacing values for negative exponential and log63

Chapter 2: Rock Mass Characterisation

normal spacing distributions. These graphs can be used to estimate population mean spacing values using the sample mean (and sample standard deviation in the log-normal case), length of scanline and type of distribution. Priest and Hudson (1981) describe a method of calculating the precision of the estimate of the population mean spacing value from the sample of size n, using standard statistical methods based on the central limit theorem. The central limit theorem states that the mean values, X , of random samples of size, n, taken from a population that follows any distribution and has some definite but unknown mean value, μx, and variance, σx2, will tend to be normally distributed 2 with a mean, μx, and a standard deviation (or standard error of the mean) of σ x / n (σ x / n) . This central limit theorem technique for analysing discontinuity spacing is particularly useful when applied to the negative exponential distribution because the mean and standard deviation are equal in this case (Priest 1993). It can be used in a number of ways. For example, it could be used to determine for a given sample, the degree of confidence (eg 90% probability) with which it can be said that the unknown population mean, μx, lies within some range of the sample mean, X . Another use would be to specify the given precision of the spacing population estimate that is required and to use this to calculate the sample size that will give the desired precision. Some drawbacks of using the central limit theorem in discontinuity analysis are that the technique is applicable only to discontinuity spacing values (Priest 1993) and that the analysis is still prone to biases such as those associated with small scanlines and orientation bias. Where rock mass discontinuity frequency is anisotropic, the estimate obtained of the range of mean total discontinuity spacing values using a given confidence limit, is only applicable to the orientation for which the sampling was carried out. 2.5.5

Discontinuity Persistence (Size) Analysis

Discontinuity persistence refers to the lateral extent or size of a discontinuity plane. In practice, the persistence of a discontinuity plane is almost always measured by the one-dimensional extent of its trace length on a sample plane. It is clear that no direct estimation of persistence is possible from borehole core, although an estimate of persistence can be made from geological inference (Hudson and Harrison 1997, Henry et al 1999). The distribution of trace lengths obtained from sampling a rock face depend to a great extent upon the degree to which length measurements are truncated and censored. The most commonly used measure of discontinuity persistence in engineering analyses is mean trace length. There are two main problems associated with estimating a mean trace length precision and accuracy (Priest and Hudson 1981). In the case of precision, it is assumed that as the number of trace length samples is increased, the sample mean trace length will tend towards the true population mean as long as there is no bias in the sampling. The truncation and 64

Chapter 2: Rock Mass Characterisation

censoring biases common in sampling trace lengths require a correction to allow an accurate measure of the trace length parameter. There has been considerable debate as to whether trace lengths have a negative exponential distribution or a log-normal distribution. The results of a number of investigations carried out on discontinuity trace length distributions are summarised in Table 2.5. It is likely that some of the differences have arisen from trace length sampling bias (Hudson and Harrison 1997). Table 2.5: Discontinuity trace length distribution and shape characteristics Reference

Trace Length

Shape

Robertson (1970) McMahon (1974)

exponential log-normal

equidimensional -

Bridges (1976) Call et al (1976)

log-normal exponential

oblong -

Baecher et al (1977) Barton (1977)

log-normal log-normal

equidimensional equidimensional

Cruden (1977) Baecher & Lanney (1978)

censored exp. log-normal or exp.

-

Herget (1982)

exponential

-

A number of techniques have been developed to estimate the mean discontinuity trace length. Some of these techniques account for the sampling biases and a number of them require some assumptions for their use. A number of the most relevant contributions are outlined below. Cruden (1977) proposed an original method to estimate the length of censored discontinuities as a function of the observed number of discontinuity end points and the observed discontinuity trace segment appearing above the sampling line (the semi-trace length). The advantage of the semi-trace length approach is that the uncensored distribution of trace length can be obtained without involving a point process of discontinuity trace centres. The problem is that semi-trace lengths are monotonic decreasing functions, insensitive to changes in the underlying trace length distribution (Villaescusa 1991). Warburton (1980) developed a stereological interpretation of discontinuity trace data. He was one of the first researchers to analyse discontinuities in three dimensions rather than two. The statistical model defines the analytical distribution of discontinuity diameters from observed trace length distributions. Unfortunately, the problem of discontinuity censoring is not addressed. Laslett (1982) developed a technique to estimate the parameters of the underlying trace length distribution from line sampling data collected in two dimensions. The technique corrects for bias incurred when incomplete observations form part of a data set. His work was limited to two dimensions. Villaescusa (1991) coupled Warburton’s stereological relationships to

65

Chapter 2: Rock Mass Characterisation

Laslett’s theory, to give a three dimensional determination of a maximum likelihood estimator of discontinuity size. Kulatilake (1988) developed a corrected equation to obtain unbiased estimates of spacing and intensity from finite scan lines. Procedures were also developed for orientation bias correction for finite size discontinuities intersecting finite size sampling domains, using a variety of discontinuity shapes (circular, rectangle, square and right angle triangle) and sampling geometries (Kulatilake et al 1990). Priest (1993) presented a graphical technique which is trace length distribution independent. This involves constructing a histogram of semi-trace lengths and drawing a best fit curve through the midpoints of the class intervals. Using the intercept of the graph, sample size and histogram class interval, an estimate of mean trace length can be obtained. The problem associated with this technique is that different interpretations of the histogram shape (by choosing different size class intervals) will result in different estimates of the underlying mean trace length. Mauldon (1998) provided new estimators of mean discontinuity trace length and density that correct for the effects of bias and censoring. A stereological estimator of mean trace length was developed which requires rectangular windows and parallel traces, and is similar to earlier methods which use a stereological methodology. Another estimator called the end-point estimator of mean trace length can be used in any convex window with variably oriented traces. This method is independent of the underlying trace length distribution in all cases and is independent of the trace orientation distribution when applied to circular sample windows. Zhang and Einstein (1998) developed a technique for estimating mean trace length from observations made using finite, circular sampling windows. It uses information on the number of end points observed in the circular sampling window and the sampling window dimensions. The advantage of the technique is that the trace lengths, the underlying distribution of trace lengths and discontinuity orientation measurements are not required. The disadvantage of the technique is the practical difficulty associated with sampling using a circular mapping window. This brief review shows that a great deal of progress has been made in increasing the accuracy of mean trace length estimates. However, little progress has been made in calculating the precision of these mean trace length estimates, perhaps because of the sampling biases inherent in the estimation of mean trace length. 2.5.6

Definition of Geotechnical or Structural Domains

Geotechnical domains are essentially regions of structural homogeneity. These regions are identified as areas of the rock mass that contain discontinuity characteristics that are more or less structurally and statistically homogeneous. If the heterogeneous ‘whole’ can be divided into homogenous parts then engineering analysis can be carried out for each design region. 66

Chapter 2: Rock Mass Characterisation

In the pre-feasibility stage of a project, before any comprehensive site investigation has been carried out, the amount of structural data available may be limited. Usually, the information collected will have concentrated on the lithologies present and the major structures contained in the region of interest. This information will be used to define the initial structural domains in the rock mass. Initial structural domain boundaries will usually be identified from: • • • •

geological boundaries between rock types; major faults; data from different levels of an underground mine; and major changes in weathering and alteration (eg unweathered rock mass and the near surface weathered zone).

In the initial analysis, it may be apparent that the rock mass discontinuity parameters found in one domain are very similar to those found in an adjacent domain. In such a case, there is no reason for maintaining the boundary between the structural domains which can be combined. Alternatively, with additional data it may become apparent that two or more zones of different rock mass discontinuity parameters may exist in the same ‘initial’ structural domain and further subdivision of the domain will be warranted. To exhibit statistical structural homogeneity a domain should have the same number of discontinuity sets and each identified discontinuity set should display similar distributions of orientation, spacing, trace length, termination index and discontinuity conditions. Quite frequently, a comprehensive set of discontinuity data containing all these parameters is unavailable. In particular, it is the number of discontinuity sets and the orientations of these sets that are used to define structural domains. The determination of homogeneous domains is often performed by visual estimation. Although subjective, visual estimation can be a reliable technique especially where an individual is familiar with the geology of the site (Bridges 1990). If visual estimation techniques are found to be inadequate, a statistically based technique may be used to cluster orientation data. An advantage of using numerical techniques is the implicit objectivity of the approach. The main disadvantage arises from the difficulty in applying statistical techniques to orientational data, particularly because of the problems associated with sampling. In order to compare two regions to establish if they are statistically structurally homogeneous, the sampling of the two regions should be identical. Even then, because of the possible poor precision in estimates of discontinuity characteristics, it is possible that identical regions may be identified as being statistically heterogeneous. 2.5.7

JK Jointstats Discontinuity Data Management System

The JK Jointstats discontinuity data management system provides the tools required to perform three important functions necessary in providing discontinuity data for subsequent engineering analysis, namely, discontinuity data input, discontinuity selection and discontinuity set definition. Once these tasks have been completed the front end program is used to initialise the appropriate analysis module. The central discontinuity database uses a Microsoft Access

67

Chapter 2: Rock Mass Characterisation

database. The widespread industry use of Microsoft Access facilitates the transfer of data to the discontinuity database and allows the data to be exported if required. Discontinuity data input Discontinuity data can be entered in different modes. Several modes are predefined and included in the program, including a number of mapping and core logging techniques. The minimum data entry mode involves only the intercept distance along the line and basic orientation data. A user-defined mode allows the user to select which discontinuity parameter fields will be used in mapping. Clearly, the data entry mode used will have great impact on the viability of the statistical calculations and models. For example, to use advanced statistical tools a measure of discontinuity trace length is required. If an attempt is made to use this module with limited discontinuity data that does not contain a measure of discontinuity trace length, a message is generated warning that such an option is unavailable. Discontinuity data can also be imported from Microsoft Excel via the MS Windows clipboard. An example of discontinuity data entry is shown in Figure 2.12. This example uses the JKMRC full scanline mapping method. Data recorded in the field can be either added directly by keyboard or if it exists in electronic form (as in field based hand held computers) it can be input via an MS Excel spreadsheet. It is possible for the user to develop mapping templates to satisfy project requirements as in the example shown in Figure 2.13. Firstly, when selecting the data collection method the ‘User Defined’ option is selected. Then the discontinuity parameters that are routinely recorded during logging or mapping are added to a blank template to create the active data sheet.

Figure 2.12: JKMRC full scanline data entry 68

Chapter 2: Rock Mass Characterisation

Figure 2.13: Creation of a user defined mapping method Discontinuity selection Discontinuity data are organised in a hierarchical manner and presented to the user in a tree view (as in MS Windows Explorer). The tree view represents an underlying database structure that stores the objects involved in discontinuity data collection in their logical relationship. These objects can be summarised as: • • • • • •

mine; mining unit (eg pit, panel, block or bench); domain (a logical grouping of discontinuity data); plane (the face from which the data were collected); scanline (the line along which data were collected); and discontinuity.

Note that cores sit at the same level as planes in the structure. This structure comprises a series of one-to-many relationships (eg one mine can contain many mining units) and referential integrity is enforced (when a scanline is deleted, all the discontinuities that belong to it are automatically removed). A typical tree view with an open menu is shown in Figure 2.14.

69

Chapter 2: Rock Mass Characterisation

Figure 2.14: Data tree view A key issue is the definition of a domain. A domain is a logical (rather than a geographic) grouping of discontinuity data. It is a region of the rock mass for which discontinuity statistics are accepted as being statistically homogenous (same orientations, size and frequency) for engineering applications. When a new mining block is created, it is automatically assigned a single new ‘hold-all’ domain. As data are collected, natural groupings will become apparent. The hold-all domain can then be split into more meaningful domains with descriptive names. Planes and cores can be cut and pasted between domains. It would be irrational to perform a statistical analysis of all the discontinuity data contained in the database, as it could contain many mines and a number of different structural domains. Discontinuities at any level from domain downward can be ‘selected’ to partition the data. Selection is achieved by clicking on the appropriate toolbar button or by toggling the selection item in the context menu. The context menu is shown after right clicking on a highlighted object and can be seen in Figure 2.14. This means that the user can toggle the selection of a given domain ‘on’ and automatically select all the discontinuities in that domain. It is possible to then manually deselect individual scanlines in the same domain to fine-tune the selection. The filter builder tool in the software package can be used to further investigate the discontinuity data for properties other than orientation (Figure 2.15). For example, it may be important to segregate the data by discontinuity type. Using such a filter the differences

70

Chapter 2: Rock Mass Characterisation

between faults, joints and veins could be assessed. Another useful exercise would be to filter the trace length parameter. Some engineering applications require only large scale discontinuities to be taken into account, so a filter could be applied to identify only those discontinuities having a discontinuity size above a set value.

Figure 2.15: Discontinuity data filter builder Since selection is not permitted across domains, any new selection that violates this rule generates a warning message. The user can then choose to replace the old selection with the new or abort the process. The selected data can then be plotted on a stereographic projection. Polar plots of discontinuity poles can be drawn in lower or upper hemisphere projection and in equal area or equal angle space. Discontinuity set definition The stereographic plot is then used to define the different discontinuity sets that exist in the rock mass. This is done by clicking on a start point, holding down the mouse button, and dragging a selection mask clockwise and outwards around the discontinuity poles. Discontinuity poles that fall inside the selection mask will turn from white to black. An option to show the Fisher distribution during set definition is provided. When the selection mask crosses the plot perimeter (ie discontinuity planes cross 90°), the mask will flip over to the ‘antipole’ region and the Fisher distribution corrected accordingly. This is equivalent to merging upper and lower hemisphere information and is valid because poles have no real termination. This transition is shown in Figure 2.16.

71

Chapter 2: Rock Mass Characterisation

Figure 2.16: Set definition with context menu Sub-horizontal sets are selected using a circular selection mask which opens from the pole. The transition from a circular to a segment selection mask is determined by the user as one of the program options. Once the set has been defined, right clicking brings up a context menu that includes an option to store the set parameters. Selecting this menu option will open the set manager form. The user is required to enter a colour for the set (so that its discontinuities can be readily distinguished in subsequent plots) and a set name. If no set name is entered, the unique ID number for that set is stored in the name field as a default. The same set manager form is also used ‘offline’ to edit or delete already established sets which are then selected from a drop-down choice box. The data set manager is shown in Figure 2.17. A variation on the data plot can be displayed along side the set manager as different sets are selected. In this mode, the plot will show only those discontinuities assigned to the current set in the stored set colour, together with the set definition mask.

72

Chapter 2: Rock Mass Characterisation

Figure 2.17: Discontinuity data set manager

Photograph gallery manager Best practice dictates that images of logged core and mapping faces are obtained during the data collection phase. The best method of storing such data is in an electronic form which allows images to be observed while the collected data are being analysed. This methodology has been incorporated into JK Jointstats by using a digital gallery manager that is linked to the discontinuity database. Images associated with the appropriate scanline or core can be easily examined using the gallery function. A number of thumbnail pictures that contain a description of the images can be viewed at once when the gallery option is selected. An example of core photographs obtained at the Newcrest Ridgeway Project is shown in Figure 2.18. The appropriate tile is simply selected to obtain a close up view of an image.

73

Chapter 2: Rock Mass Characterisation

Figure 2.18: Discontinuity photograph gallery Once discontinuities have been selected and assigned to sets, the program can run simple statistical calculations on the information. These are shown on a multi-tabbed form containing tables and histogram plots. Any plot can be copied to the clipboard, printed or dumped to file. Any table can be transferred to MS Excel via the clipboard. A number of simple statistics are available as outlined below. Rock mass spacing (average discontinuity spacing in the rock mass) The discontinuity spacing for all the selected discontinuities, whether assigned to sets or not, is displayed on the second tab of the Basic Statistics module (see Figure 2.19). This gives an impression of the overall discontinuity spacing in the rock mass and an appreciation of the potential in situ block size. The average fracture frequency in the rock mass is also estimated.

74

Chapter 2: Rock Mass Characterisation

Figure 2.19: Rock mass discontinuity spacing

Corrected discontinuity spacing by set Given that the orientations of the discontinuity planes are known and the mean orientation of each set has been calculated, the apparent (measured) discontinuity spacing can be corrected for relative orientation and plotted for each set. This apparent spacing can then be converted to a true or corrected spacing. The statistics of the discontinuity set spacings are calculated and the results are shown graphically as illustrated in Figure 2.20. Using the pull-down set selector near the top of the form (see Figure 2.20), it is possible to review and compare the results for the discontinuity sets selected in the analysis.

75

Chapter 2: Rock Mass Characterisation

Figure 2.20: Corrected discontinuity spacing per set

Trace lengths by set and for random discontinuities Statistics for the recorded trace lengths can be displayed and plotted for each set and for those discontinuities selected but unassigned to sets (termed ‘random’ discontinuities). An example for random discontinuities is shown in Figure 2.21. The plot takes the form of a stacked histogram, the elements of which are designed to show a qualitative assessment of ‘confidence’. This is necessary because discontinuity trace lengths can be censored by the observation window limits at either the top or bottom (shown in grey shades) or at both the top and bottom (shown in black). Clearly, the censored discontinuities will tend to occur at the greater trace lengths (depending on the window limits) so the data are less meaningful towards the right of the plot. However, this remains potentially useful information because it is known that such trace lengths fall outside the lower bins. Thus, it is important to show such data as long as some impression of confidence is included.

76

Chapter 2: Rock Mass Characterisation

Figure 2.21: Trace length statistics for ‘random’ discontinuities In the Basic Statistics selection the censoring classes provide a qualitative measure of confidence in the trace length statistics. The information on the censoring class of the discontinuities is vital when the advanced statistics option is selected as it is used in the estimation of the mean size parameter of the discontinuities. It is only by utilising the advanced option that a quantitative measure of confidence on discontinuity parameters can be obtained.

2.6 2.6.1

SIMULATION OF ROCK MASS GEOMETRY Introduction

The simulation of a rock mass geometry involves the construction, with the aid of a computer, of a graphical representation of the rock joints within the rock mass. The graphical representation may be a 2-D section through the rock mass, effectively showing the traces of the joints on a face, or it may be a full 3-D model in which the assembly of joints may be viewed from various angles or the observer may ‘fly’ through the rock mass, inspecting the joints and their intersections. Simulations of a rock mass must be based on some conceptual model of the rock joints and the manner in which they occur in the rock mass. The conceptual model effectively provides the rules according to which the joints are placed into 3-D space. Thus the geometric features of

77

Chapter 2: Rock Mass Characterisation

the simulation and any geomechanical consequences that they may have are a function of the model and will be only as valid as the model itself. A number of models may be conceived of as being plausible (Dershowitz 1993). The extent to which any one model actually mimics nature must be established by the collection of data and comparison of the data with the predictions of the model after it has been matched to the data as well as possible. There are a number of reasons why it is useful to be able to simulate a rock mass geometry. Firstly, it is an aid to the visualisation of the joints in the rock mass. Given the belief or even reasonable evidence that there are particular joint sets in the rock mass at particular densities and with particular size and orientation distributions, the use of the computer to view the interrelationships of the joints allows the engineer to confirm mental constructions of the joints. The interaction of joint sets in a rock mass is a complex process. The complexity derives not so much from the model as from the fact that there are many joints of many possible orientations and sizes. And, since it is generally taken that the jointing is a stochastic phenomenon, the possibilities for joint interaction are effectively infinite. The interaction of joints in the rock mass controls the formation of rock blocks. The existence of discrete, fully formed blocks is of vital interest for the design of reinforcement or for the design of block caves. The in situ fragmentation (the size distribution and volumetric concentration of formed blocks prior to the commencement of mining) is important not only for assessing the stability of the rock mass once mining commences but also for the purpose of estimating the size distribution of the ore at the draw points following secondary fragmentation. The 3-D network of joints also defines pathways for fluid flow through the rock mass. Simulation of discontinuities in a rock mass, in general, is a computational problem of varying tractability; the level of tractability depends on the rules chosen to define the way in which the discontinuities occur. The random disk model is perhaps the simplest of all models as all disks are independent. The more complex the rules, the more difficult is the program simulation and the longer the execution times to complete a simulation in a given rock mass volume. 2.6.2

Approaches to Discontinuity Modelling

Introduction Models of rock discontinuity networks are a fairly recent development in rock engineering. The first conceptual models such as Snow’s (1965) orthogonal model were developed as relatively simple tools for hydrological modelling. More advanced methods such as the Poisson location models (Baecher et al 1977), tessellation models (Veneziano 1978), hierarchical methods (Lee et al 1990), fractal approaches (Barton and Larson 1985) and geostatistical methods (Gervais et al 1995), have allowed more realistic representations of in situ discontinuity network geometries to be created. Some of these approaches are outlined below.

78

Chapter 2: Rock Mass Characterisation

Random disk (Poisson location) discontinuity models The fundamental feature of the random disc model (Baecher et al 1977, Warburton 1980, Villaescusa 1991, Hadjigeorgiou et al 1998) is the assumption of circular or elliptical discontinuities. The size of circular discontinuities is defined completely by a single parameter, the discontinuity radius Rj. The discontinuity radius may be defined deterministically as a constant for all discontinuities, or stochastically by a distribution of radii fr(Rj). Appropriate distributional forms for discontinuity radius include the exponential and lognormal distributions, both of which produce lognormal distributions of discontinuity trace length (Baecher 1983). Discontinuity location may be defined by a regular (deterministic) pattern or by a stochastic process. The simplest stochastic assumption is a Poisson process, in which discontinuity centres are located randomly and uniformly in space. Discontinuity orientations may be defined by any orientation distribution, or by a constant orientation. As a result of the discontinuity location, shape and size process of the Baecher model, discontinuities intersect each other and terminate in intact rock (Dershowitz and Einstein 1988). An example of this model is shown in Figure 2.22.

Figure 2.22: Random disk model (Dershowitz and Einstein 1988) The mutual independence of discontinuities is the biggest disadvantage of the random disk model, in that the discontinuity termination often seen in rock mass exposures cannot be modelled. Interestingly though, this independence of discontinuities is also the source of the greatest strength of the random disk model. It permits the application of a number of statistical techniques in conjunction with the rock mass model which rely upon the independence of

79

Chapter 2: Rock Mass Characterisation

elements. This allows a link to be developed between two dimensional discontinuity properties measured in the field (spacing or frequency and trace length) and those three dimensional discontinuity properties modelled (intensity and size). Although the model was initially developed using circular disks, there is no reason why other shapes cannot be used to represent discontinuities. The Poisson location seeding and orientation process remains unchanged. However, the shape of the discontinuity does affect the analytical techniques developed to relate trace length properties with the discontinuity size properties but this can be overcome by using a forward modelling technique. Random coplanar polygon discontinuity models Random coplanar polygon discontinuity models use a tessellation approach to model rock mass discontinuities. A tessellation is a division of a plane into polygons, or of space into polyhedra. When discontinuities are sufficiently connected to produce completely defined rock blocks, a tessellation approach may be appropriate. The generation of a Veneziano (1978) rock discontinuity system model requires three consecutive stochastic processes: • • •

discontinuity planes generated as infinite Poisson planes, distributed in space by a uniform distribution with any distribution of orientation allowed; a Poisson line process on each discontinuity plane divides discontinuity planes into polygonal regions; and a portion, PA, of these polygons is randomly marked as jointed, while the remainder are marked as intact rock where PA corresponds to persistence.

Discontinuities are modelled using polygonal shapes, and discontinuity sizes are defined by the intensity of the Poisson line process and the proportion of the polygons marked as discontinuities. The Veneziano model resembles the Baecher model, except that discontinuities are represented by coplanar line segments rather than independent circles (lines in two dimensions). In the Veneziano model, an independent Poisson line process defines discontinuities in each plane. This process means that discontinuities in one plane are completely independent of discontinuities in adjacent planes, so termination at another discontinuity cannot occur except by chance (Dershowitz and Einstein 1988). One of the problems associated with the Veneziano model is in the production of in situ blocks. Discontinuities are defined as Poisson lines on previously defined Poisson planes. Intersections between discontinuities on different discontinuity planes, therefore, do not often match discontinuity edges. Rock blocks can be created with Veneziano models if the discontinuities are 100% persistent and unbounded but the resulting infinite persistent discontinuity planes provide an unrealistic assumption for the modelling a rock mass.

80

Chapter 2: Rock Mass Characterisation

Figure 2.23: Dershowitz random coplanar polygon discontinuity model (Dershowitz and Einstein 1988)

The development of the Dershowitz (1984) model overcame some of the problems of the Veneziano model. Dershowitz locates the discontinuity planes in space in a manner similar to that used in the Veneziano model but the polygons are formed by the intersections of the Poissonian flats (see Figure 2.23). A portion of the polygons are marked as broken and the remainder as intact rock. This model improves the previous one, since all the discontinuities can terminate against discontinuity edges. Although this model can define distinct blocks at any scale (block faces are either completely broken or intact), an increase in the Poissonian flat density may produce a problematically large number of small polygons (Dershowitz and Einstein 1988). Dershowitz has further advanced tessellation models to develop porosity analyses of fractured rock masses (Dershowitz 1993). An advantage of the Dershowitz (1984) model is that distinct rock blocks are defined. Another advantage is that by using a flexible orientation distribution a variety of polygonal discontinuity shapes and polyhedral block shapes can be modelled. It is possible to create discontinuities terminating in intact rock (and hence intact rock bridges) by modelling with a virtual discontinuity set which has zero persistence. Random coplanar discontinuity polygon models do not model the inherent statistical properties of discontinuities (size, shape, intensity and orientation) which is a significant disadvantage. Another disadvantage outlined by Dershowitz and Einstein (1988), is the fact that polygonal block face sizes are controlled by the intensity of intersecting discontinuity plane processes. As

81

Chapter 2: Rock Mass Characterisation

the intensity of the plane process increases, the number of intersecting lines on each plane increases, and therefore the size of the polygons defined by the line decreases. If joints are defined as a constant percentage of each plane, the increase in plane process intensity results in a larger number of smaller polygons. Fractal based discontinuity models Fractal geometry is a way of quantifying the spatial or temporal dispersion of a quantity, and how this dispersion may change with the scale of observation. Its attraction for discontinuity pattern characterisation lies in its simplicity, the less rigorous data requirements (than geostatistical approaches) and its correspondence to certain geometric features in natural discontinuity patterns. An early application of a fractal based discontinuity model was a two dimensional model of jointing at the proposed nuclear waste repository at Yucca Mountain, USA (Barton et al 1985). One of the characteristic features of fractal models is that large discontinuities are not often found near other large discontinuities; rather, they are spaced at large intervals, so that, on average, the size of the fracture is proportional to the spacing. This autocorrelation of discontinuity trace length would be picked up using geostatistical models of trace length semivariograms. Empirical evidence does not unequivocally support the existence of autocorrelation, so to use a model whose main strength is the ability to model autocorrelation of discontinuity properties may be pre-emptive until further study in this field is undertaken. The fractal nature of a discontinuity network is examined using discontinuity trace maps. The reliance on trace maps to characterise the fractal nature of the discontinuity network may limit their use in cave mine engineering, due to the extensive use of core logging as a source of discontinuity characterisation information. Although promising, the application of fractal geometry for engineering design purposes is largely untested. For example, no verification exists that fracture networks derived from fractals produce model input that predicts the behaviour of a flow experiment any more accurately than other methods (La Pointe 1993). Hierarchical discontinuity models In hierarchical models the discontinuity sets are described and modelled in hierarchical order to account for dependencies among discontinuities of the same set or of different sets. The Hierarchical Fracture Trace Model of Lee et al (1990) was one of the earliest hierarchical models. The sequential generation and correlation of discontinuity sets corresponds to what happens in nature. The main features of the model are: •

The spatial variation of trace density is represented through a double stochastic point process. This class of procedure is quite versatile and is especially appropriate when the variation of trace density is due to external factors (eg state of stress, rock strength).

82

Chapter 2: Rock Mass Characterisation

• •

•

A new method, based on maximum likelihood theory, is presented for the unbiased estimation of trace length distribution using outcrop data. When several trace sets are present, these sets are analysed sequentially, according to a hierarchical (ideally, chronological) order. Each set is represented through a conditional stochastic process, conditioning being with regard to the lower-order sets. This is perhaps the most innovative feature of the model. Methods from multivariate point processes are adapted to the estimation and validation of line-segment processes in tessellation.

The model contains several interesting features that would be desirable in a rock mass model for cave mine engineering. Unfortunately, the model would have to undergo further development before becoming a useful caving rock mass characterisation tool. The advances required would include extension of the model into three dimensions and the analysis of patterns with more than two trace sets for structurally complex rock masses. A recent hierarchical model specifically designed to model real geological fracturing processes, is the MIT geologic stochastic fracture model (Meyer et al 1999). The model is hierarchical since the fractures produced are grouped into hierarchically related fracture sets. It uses tessellated Poisson planes which have been subdivided into fractured and unfractured rock, and uses the geological history of the discontinuities to recreate the geometric network seen in the field. Although applied to large scale discontinuities, namely crustal scale shears and large scale faulting, it nevertheless applies some novel techniques to obtain a good representation of fracture geometry. In the example discussed by Meyer et al (1999), four stages of faulting are modelled progressing from a simple to a compound arrangement of strike-slip fault zones. Probability is used to determine whether a new fault crosses through a previously existing fault when an intersection occurs. Geostatistical discontinuity models In reviewing traditional stochastic discontinuity geometry models, Gervais et al (1995) found that discontinuity patterns observed in real rock masses were not adequately recreated in the models because of the variable discontinuity density and clustering of discontinuities occurring in nature. They proposed a geostatistical model of discontinuity geometry. This model features a hierarchical model of discontinuity networks, utilising a statistical characterisation and a geostatistical approach. Semi-variograms were developed to describe the spatial behaviour of the regionalised variables of the discontinuity network, and thus of the underlying random process. Semi-variograms of cumulated length of discontinuity traces per square meter were computed to study the spatial correlation of discontinuity density. For each discontinuity set, the variogram shows a structure along the direction of the mean orientation of the set. Spatial correlation of discontinuity intensity exists in this orientation. This is because well defined 83

Chapter 2: Rock Mass Characterisation

corridors of discontinuities line up with each discontinuity set. Variograms constructed perpendicularly to the direction of the mean of the set, show a pure nugget effect with no structure. To model the geometry of the discontinuity network, the first discontinuity set was generated with the classical method (random distribution of discontinuity centres, log-normal distribution of trace lengths), after which the other sets were generated successively and conditionally upon the previous ones, in order to reproduce the geometrical relations between sets and the discontinuity termination types observed in the field (Gervais et al 1995). The Gervais et al (1995) model was developed on discontinuity trace sets of a bedding plane in a limestone quarry, where extensive discontinuity trace maps could be used as the basis of discontinuity characterisation. As a result, the study is highly detailed but entirely two dimensional. Although geostatistics can be an extremely useful tool in characterising the spatial variability of discontinuity intensity, it must be applied in three dimensions to allow for its application in underground mining. However, it is even more important to realise the limits imposed by the generally poor sampling of rock masses in cave mining operations, which will rule out the possibility of developing accurate semi-variograms of discontinuity set parameters. Forward modelling techniques Forward modelling is a method of obtaining parameters for constructing three-dimensional discrete models of discontinuity systems. In these models, each discontinuity is represented individually and has unique properties (as in the Baecher random disk model). The properties of the model are derived from iterative conditioning to observed data (La Pointe et al 1983). The forward modelling approach is illustrated in Figure 2.24. The method used to collect the structural data is simulated together with the biases described previously. This results in a simulated set of field data which can be compared directly to the field measurements. The goodness of fit between the field and simulated data can be evaluated visually, by statistical comparison, or by the use of statistical tests such as the χ 2 and Kolmogorov-Smirnov tests (Dershowitz 1995). Based on the results derived from the comparison, the statistical description of discontinuities (orientation, size, shape, spatial distribution) can be modified and the process repeated to obtain a satisfactory match between field and simulated data. Knowing that modelled discontinuity measurements match the in situ discontinuity measurements provides greater confidence in the discontinuity model analyses. Dershowitz (1995) also illustrates other advantages of forward modelling approaches over mathematical approaches such as that used by Villaescusa (1991). The most important is the ability to directly account for known bias, censoring and truncation processes. The forward modelling approach directly simulates the methods of data collection, and therefore automatically accounts for these biases.

84

Chapter 2: Rock Mass Characterisation

Assumed Geometric Properties Similate process

of measurement used in-situ

Simulated In-situ Measurements

Adjust assumptions for better match

Generate additional alternative conceptual models if appropriate

In-situ Measurements

Compare Simulated and Field Measurements Good match

Accept Possible Conceptual Model Geometric Properties

Figure 2.24: Forward modelling approach (adapted from Dershowitz 1995) 2.6.3

The Development of the JKMRC 3-D Discontinuity Model

This account of the JKMRC 3-D discontinuity model developed as part of the ICS Stage I, is based on the reports of Lyman (2000) and Harries (2001). The simplest model of rock joints that does not have infinitely persistent joints is the random disk model which is a Boolean Random Set (BRS) model. The term Boolean set is used because a set element is or is not present at a point in space. Set elements are placed entirely independently at Poisson points in space and are allowed to overlap or interpenetrate freely. Consider the simple example of spheres placed randomly in space. Having first decided on a volume of space in which the random set is to be constructed, the space is populated with Poisson points of a chosen spatial density, or number of points per unit volume. A set element is then placed at each one of these points. Assume that the volume is a cube 10 m on each side and the spheres have a constant radius of 1 m as shown in Figure 2.25.

85

Chapter 2: Rock Mass Characterisation

Figure 2.25: BRS of spheres in a cubical volume If the simulation is to be entirely valid within the cube, the sphere centres have to be located in a volume slightly larger than the cube to avoid edge effects. In fact, to consider sphere centres, the volume is found by extending the cube by one sphere diameter in all directions. The easiest method in this case is simply to consider a cube 12 m on edge, centred on the 10 m cube. By definition, if Poisson points have a spatial density of ρ (points per cubic meter), then the probability of finding a point in a small volume dV at an arbitrary location is ρ dV . The expected number of points to be found in a finite volume V is

nV = ρV

(2.2)

and the actual number found in V follows a Poisson distribution with expected value nV . The numbers of points occurring in different volumes in the space are mutually independent. If the spatial density chosen is say 2 m-3, the correct way to put the points into the 12 m cube is to first generate a Poisson random number having the expected value

nV = 2 ×12 = 24 Say this random number turns out to be 29. Choosing 29 triples of random numbers that follow a uniform distribution between 0 and 12 provides the coordinates of the points. The random number generator used for this has to be a good one (see Press et al 1992).

86

Chapter 2: Rock Mass Characterisation

The centres of the spheres are placed on the points. This is now a realisation of the random set. Any number of independent realisations can be generated and the extent to which the spheres make contact can be investigated for each realisation and an average measure of the number of touching spheres determined by averaging over many realisations. The set elements used in the above example had a constant diameter; it would be quite legitimate to define the sphere diameters according to some statistical distribution. The distribution could be a discrete one with a particular average or expected proportions of, say, four different diameters. Alternatively, the distribution could be a continuous one quantified by a probability density function g ( D ) such that the probability of finding a sphere of diameter D to D + dD is g ( D ) dD . To generate a realisation of spheres whose diameters follow such a distribution, the probability distribution function corresponding to g ( D ) is calculated, providing the probability that a sphere has a diameter smaller than a given value. This probability is

G ( D′ ) = Pr { D < D′} D′

=

(2.3)

∫ g ( x ) dx 0

The function G ( D ) must rise monotonically from 0 to 1. Therefore, a sphere of diameter Di can be chosen from g ( D ) by choosing a random number ri uniformly distributed between 0 and 1 and using the function G ( D ) as shown in Figure 2.26. To make the random set with the sphere diameter probability density g ( D ) , simply choose a sphere diameter as illustrated in Figure 2.26 before placing the point in space. If disks are used as the random set elements, an additional factor comes into consideration, namely the disk orientation. The sphere is a completely symmetric object and so has no detectable orientation. A very thin disk has only one axis of symmetry and so needs two numbers to specify its orientation.

87

Chapter 2: Rock Mass Characterisation

G ( D)

1

ri

0

D Di

Figure 2.26: Illustration of a probability distribution function and the sampling of values from a specific density function

As has been noted previously, the orientation distribution most commonly used for jointing is the Fisher distribution (Fisher 1953). The Fisher distribution is the counterpart of the normal (Gaussian) distribution that is suitable for the topology of the sphere. Figure 2.27 shows a unit sphere and the two angles necessary to specify a point on its surface. z

δ

N θ

y

θ θ

x

E

Figure 2.27: Unit vector from origin to surface of unit sphere showing definition of angles A unit vector from the origin to a point on the unit sphere can be defined by the angle θ between the North-directed y-axis, measured clockwise when looking from above and the angle δ between the z-axis and the vector. The vector is then defined as

u = sin δ sin θ ˆi + sin δ cosθ ˆj + cos δ kˆ

88

−π < θ ≤ π 0 ≤δ ≤π

(2.4)

Chapter 2: Rock Mass Characterisation

The orientation of a disk is completely defined by the direction of its unit normal vector. Equation 2.4 can be used to define the normal. In fact, because a disk also possesses a plane of symmetry, it can define either an upward- or a downward-pointing vector. For joint orientations, a convention of downward-pointing vectors is adopted and these vectors are represented on the lower hemisphere stereoplot. The Fisher distribution is defined on the entire sphere in terms of a mean vector direction and a dispersion value K. The probability density function for the Fisher distribution is

φ Fisher (δ ,θ , μ, K ) d δ dθ =

K exp ( K n (δ ,θ ) ⋅ μ ) sin δ dδ dθ 4π sinh K

where n (δ ,θ ) is a vector in a direction defined by angles

(δ ,θ )

μ

(2.5)

on the unit sphere and

μ (δ μ ,θ μ ) is the average direction of the vectors defined by angles (δ μ ,θ μ ) . Note that n (δ ,θ ) ⋅ μ (δ μ ,θ μ ) = cos δ n

(2.6)

or the cosine of the angle between the vector n and the mean direction

μ . It is very important

to recognise that this distribution is defined on the entire sphere and not on just the upper or lower hemisphere. The vectors are illustrated in Figure 2.28.

Figure 2.28: Vector n, a sample from a Fisher distribution with mean direction

μ

It has been shown by Lyman (2000) that the expected value and variance of cos δ n for the Fisher distribution are given by

E {cos δ n } = coth K −

1 K

(2.7)

89

Chapter 2: Rock Mass Characterisation

Var {cos δ n } = 1 − coth 2 K +

1 K2

(2.8)

For large K, coth K → 1 quite rapidly so that

Var {cos δ n } →

1 K2

(2.9)

Equation 2.9 shows how K provides a measure of spread of the distribution. The Fisher distribution is plotted in Figure 2.29. 7

6 100

Density

5

4 50 3 20 2

10 5

1

1 0 0

30

60

90

120

150

180

Angle δ [degrees]

Figure 2.29: The Fisher probability density function for a series of values of the Fisher constant K If the Fisher distribution is modified to account for a non-symmetric distribution of the normal vector directions, it can be simply multiplied by an appropriate density function in θ to arrive at, for example,

φ (δ ,θ , δ n ,θ n ) dθ dδ = f (θ n )

K exp ( K cos δ n ) sin δ dδ dθ 2sinh K

(2.10)

A completely general form of the orientation density function is

φ (δ ,θ , δ n ,θ n ) dδ dθ = b (θ n , δ n )

sin δ dδ dθ 4π

90

(2.11)

Chapter 2: Rock Mass Characterisation

where b (θ , δ ) = 1 for a uniform distribution on the sphere. To sample from the Fisher distribution, the distribution function for the Fisher constant is required. If r is a random number uniformly distributed on the interval 0 ≤ r ≤ 1 , then

cos δ n =

1 ⎡ ln r sinh K + K ⎢⎣

( r sinh K )

2

+ 1⎤ ⎦⎥

(2.12)

θ n must also be specified, and any convenient reference direction from which to measure θ n may be chosen. A value of θ n can be generated as

is a Fisher deviate. The angle

θ n = π ( 2r − 1)

(2.13)

where r is a second uniform random number. Choosing the plane defined by vertical vector as a reference for

θn ,

μ and a purely

the Fisher normal can be rotated back to the original

coordinates by the calculation

⎡ n1 ⎤ ⎡ cos θ μ ⎢ n ⎥ = ⎢ − sin θ μ ⎢ 2⎥ ⎢ ⎢⎣ n3 ⎥⎦ ⎢⎣ 0

cos δ μ sin θ μ cos δ μ cos θ μ − sin δ μ

sin δ μ sin θ μ ⎤ ⎡ sin δ n sin θ n ⎤ ⎥ sin δ μ cos θ μ ⎥ ⎢⎢sin δ n cosθ n ⎥⎥ cos δ μ ⎥⎦ ⎢⎣ cos δ n ⎥⎦

(2.14)

These relatively simple calculations are all that is needed to generate a set of vectors coming from a Fisher distribution. Note that scanline orientation bias is not included. These procedures are sufficient to create a BRS realisation of random disks following a Fisher distribution in a chosen volume of space. The line of intersection between a plane within the simulation volume and any one of the disks may then be determined. The intersections between a line representing the axis of a core and the set of disks generated may also be found. The basic BRS model can also be varied to use shapes other then disks for the joints if desired. Lyman (2000) has applied maximum likelihood theory to the BRS model to produce a number of important results. Lyman’s approach is statistically rigorous, allowing all geometric and censoring biases to be accounted for exactly in establishing values of parameters such as joint size and density. The likelihood distribution functions for joint parameters may be determined by using fitted parameter values and trial data sets.

91

Chapter 2: Rock Mass Characterisation

The direct use of the likelihood distribution function permits confidence intervals to be established for estimates of joint parameters. The application of this modelling approach has allowed the impact of additional scanline sampling on parameter confidence intervals to be tested and the impact of data quantity and quality on the parameters used in cave design to be assessed (Harries 2001). In principle, the approach is also capable of being used to develop an automatic, statistically based clustering method to sort joint orientation data into sets (Lyman 2000). 2.6.4

The JKMRC Hierarchical Model of Discontinuity Network Geometry

The geometrical properties of the rock fractures that have been modelled in the JKMRC 3-D rock fracture model (see Section 2.6.3) include: • • • •

discontinuity orientation; discontinuity shape; discontinuity size; and discontinuity volumetric density.

The first three parameters wholly describe the geometries of individual discontinuities. Discontinuity volumetric density relates to the density or intensity of fracturing of a given set. However, these are not the only discontinuity parameters needed to describe fully the geometry of the discontinuity network. There are also discontinuity network parameters such as termination style, degree of interconnection and hierarchy seen in the network. Figure 2.30 illustrates some of these network parameters schematically.

Case A

Case B

Figure 2.30: Influence of fracture network parameters on geometry The two fracture trace maps shown in Figure 2.30 have identical individual fracture geometry statistics. The fracture sets in both cases have matching orientations and sizes, only their locations differ. However, it can be seen clearly that the fracture network geometries for the two cases are quite different. The distributions of primary fragments or blocks defined by the 92

Chapter 2: Rock Mass Characterisation

fracture traces are very different. In Case A, the fractures have been created as independent events with no mutual interaction (as in the JKMRC 3-D model). Case B has a definite hierarchy in that the fractures do interact with each other. These interacting traces more often fully define polygons in the resulting simulation. Case B illustrates the influence of fracture termination on the resulting discontinuity network geometry. The statistical procedures used for estimating fracture parameters in the 3-D model provide the starting point for the hierarchical modelling exercise. However, given the statistical (and independent) nature of the rock discontinuity parameters, an infinite number of rock discontinuity geometries are possible. This may result in the creation of a rock mass model in which the fracture parameters are statistically correct when compared to those measured in the field but the modelled fracture network geometry may be vastly different from that seen in the field. For the construction of a more realistic model of the rock discontinuity network, additional rock mass parameters need to be included in the modelling exercise. A review of the literature suggests that one of the main determinants of the geometry of discontinuities within the rock mass is the timing and termination of discontinuities (Harries 2001). The geological history of discontinuities can be estimated from knowledge of the structural geology and, in particular, any cross cutting relationships observed in the rock mass discontinuities. A blocky pattern that appears to be defining in situ blocks can be seen in the discontinuity network shown in Figure 2.31. This is partly due to the different timings of discontinuity formation and their terminations upon one another. In Figure 2.31 many of the discontinuities terminate on other discontinuities. It is unlikely that the geometry of the rock mass could be accurately recreated without modelling the termination seen in the rock mass. A number of engineering applications rely on a representative determination and modelling of the rock mass discontinuity geometry. These include the determination of in situ block size and the hydraulic conductivity of a rock mass. This section will confine itself to the hierarchical modelling of rock mass discontinuities to better simulate the rock mass geometry. The use of this modelled discontinuity network geometry in the estimation of primary fragmentation will be discussed in Chapter 4.

93

Chapter 2: Rock Mass Characterisation

Photograph

Digitised discontinuity trace map

Figure 2.31: Discontinuity network, Kangaroo Point, Brisbane, Australia (after Harries 2001) Discontinuity termination statistics The International Society of Rock Mechanics (Brown 1981) has suggested that termination data be presented as a termination index, Tr, calculated as

Tr =

100 Nr Nr + No + N j

(2.15)

where Nr is the number of fracture endpoints seen terminating in intact rock, Nj is the number of fracture endpoints seen terminating on another joint and No is the number of fracture endpoints not seen terminating (obscured). This termination index is a useful measure of discontinuity termination. A high Tr value suggests that the rock mass will contain many intact rock bridges, rather than being made up of discrete blocks. For the geometry to be accurately reproduced, a modelled rock mass must have a termination index consistent with that measured in the field. However, the Tr measure cannot account for fractures crossing other fractures (referred to as ‘X’ intersections). This factor has to be taken into account during the modelling exercise if a realistic rock mass geometry is to be recreated. Given that the intersection of fractures is a statistical event, we need to calculate a probability of fracture termination (and consequently a probability of the fracture crossing 94

Chapter 2: Rock Mass Characterisation

through the pre-existing discontinuity). The probability of fracture termination needs to be developed using the data collected on fracture termination and the number of crossing fractures seen. Discontinuity termination probability The number of fractures terminating in intact rock, the number terminating on other fractures and statistics on fracture crossings are now routinely included in the JKMRC rock mass characterisation methodology (Harries 2001). Using this information for a given discontinuity set s, the fracture termination probability is calculated using n

Pr = s t

∑T i =1

n

∑ (T i =1

i

i

(2.16)

+ Xi )

where, Prts is the probability of termination for the given fracture set s, n is the number of fractures in a given fracture set s, Ti is the number of observed terminations (referred to as T intersections because of the pattern they form) for each fracture in the set (0, 1 or 2) and Xi is the number of crossing X intersections seen for each fracture in the set (0, 1, 2, …, ∞). The probability of a discontinuity crossing the pre-existing discontinuity is simply

Prxs = 1 − Prts

(2.17)

where, Prxs is the probability of fracture crossing (X intersection) for the given fracture set s. Since the estimate of discontinuity termination probability is based on observations of fracture endpoints, the estimate will be biased by the size of the sampling window within which the measurements were taken. This geometrical bias is similar to the bias in the trace length distribution discussed in Section 2.5.5. However, because of the iterative nature of the hierarchical model, the initial estimate of termination probability can be modified to better fit discontinuity observations made in the field. Overview of the hierarchical model The hierarchical rock mass model developed by Harries (2001) implements the modelling of discontinuity ordering and termination. This sequential generation and correlation of fracture sets is intended to correspond to what happens in nature. In developing the fracture termination model, four key aims were:

95

Chapter 2: Rock Mass Characterisation

• • • •

modelling the structural history of the rock mass by simulating the ordering of discontinuities which make up the rock fracture network; modelling both X and T type discontinuity intersections; the modelling method is to be generic enough to allow all structural types of rock mass to be modelled successfully; and to limit the change in the discontinuity parameter set statistics (number, size and orientation) to an insignificant level, as best estimates of these parameters have been obtained from the 3-D statistical modelling.

Achieving these aims involves two main processes. Firstly, ordering the fractures to satisfy structural history considerations and secondly, modelling the fractures sequentially, checking for intersections between the fractures. Where there is an intersection, the associated probability of termination of the discontinuity should be used to assess whether to terminate the fracture or not. Model inputs The hierarchical model requires a number of inputs for it to perform in a satisfactory manner. The computer program prompts the user for these inputs: • • • • •

the number of fracture sets in the model; the number of different structural events to be modelled (separate geological events of discontinuity creation); sampling plane dimensions (height and width); geological history ordering of the different discontinuity sets; and probability of termination for each discontinuity set.

Initial data structure The initial data concerning the modelled discontinuities is derived from the output of the JKMRC 3-D discontinuity model (see Section 2.6.3). An example of the visual output of the 3D model is shown in Figure 2.32. A 20 m long and 10 m high sampling plane is depicted with the discontinuities that intersect the plane. Two different discontinuity sets are modelled in this simple example. One fracture set is steeply dipping with large discontinuities whose orientations are roughly perpendicular to the sample plane. The second fracture set modelled is oriented horizontally with a smaller mean size and a higher volumetric density than the vertical set. In reality, most rock masses have three or more fracture sets with a far greater dispersion about the mean orientation of the fractures. It often becomes difficult to develop an understanding of the rock mass by viewing such 3-D realizations when the complexity of the rock discontinuity geometry approaches reality.

96

Chapter 2: Rock Mass Characterisation

Figure 2.32: The 3-D rock mass model showing two example fracture sets (Harries 2001)

The output used in the hierarchical model is the discontinuity information contained in the 2-D sections. Figure 2.33 shows a visual example of the output plane. This plane is the same 20 by 10 m section modelled in the 3-D simulation shown in Figure 2.32. The two discontinuity sets contained in the model can be clearly identified. The flat lying horizontal line 2 m above the floor of the simulation represents the location of the scanline from which the discontinuity information was collected.

2m Figure 2.33: The 2-D sample plane showing the same two fracture sets as in Figure 2.32 97

Chapter 2: Rock Mass Characterisation

There are no limitations to the number of planes that can be modelled in the 3-D model. For the sake of visual clarity only one plane is shown in Figure 2.32. Usually multiple planes are modelled with different orientations. Some plane orientations will be sub-parallel to the modelled discontinuity sets and consequently these sets will be poorly sampled. By using a large number of differently oriented planes, this geometric sampling bias can be reduced. Every discontinuity intersecting the sample plane contains information pertaining to the fracture set number, the trace length, the end point vertices, and information concerning the start and end termination codes. These termination codes are initially set to zero (representing intact rock termination) as the 3-D model does not consider termination. The fracture endpoint termination codes are used later in the termination modelling. Ordering of discontinuities to mimic structural history An organisational chart illustrating the hierarchical modelling process is shown in Figure 2.34. The first step involves ordering the fractures to model the structural history of the rock mass. Unterminated and Unordered Fractures Structural History and Ordering of Fractures Calculate Fracture Trace Length Statistics Model Sequential Placement of Fractures

Next Fracture

Does Fracture Cross Pre-Existing Fracture YES

NO

Review Probability of Termination

Fracture Terminates

NO All Fractures Modelled

No Termination

Modify Initial Fracture Sizes or Probability of Termination

YES Conserve Tracelength Where Possible Change Endpoint Coordinates and Endpoint Codes

Calculate Tracelength Statistics - Acceptable Fit NO

YES

Calculate Termination Statistics - Acceptable Fit NO

YES

Prepare for Primary Fragmentation Modelling and Calculate Engineering Indices

Figure 2.34: Organisational chart of the hierarchical modelling (Harries 2001) 98

Chapter 2: Rock Mass Characterisation

The hierarchical model uses field structural geological data to designate the historical order of formation of the different discontinuity sets (see Harries 2001). Where no hierarchical structure can be identified, a simple ordering by joint length is required. The longest fractures are more likely to be the earlier fractures formed (Hudson and Cosgrove 1997) and so are modelled first. Where multiple structural events are being modelled, a more intricate sorting routine is required. A number of procedures have been devised to order the fractures firstly by structural history and then by length. Fractures are ordered by length from the longest to the shortest within the different structural history groupings. After the fractures have been sorted in this manner, the discontinuity data are ready for the termination modelling. Fracture termination model The fracture termination model uses geometric processes to model mechanical processes that occur in nature, namely fracture intersection and termination. After the ordering of the fractures, the next step is to determine the trace length statistics. The trace length mean, standard deviation and sum are calculated. These are used later to ensure that the termination model does not significantly alter the trace length statistics. Fractures are then placed in the order designated by the structural history and length considerations. The first fracture placed has no chance of intersecting another fracture because it is the only fracture on the plane. As subsequent fractures are placed into the simulated plane they are checked to see if they intersect any previously placed discontinuities. Since the fractures are modelled as planar traces, intersections can be checked using plane equations and by checking the limits of the discontinuity traces. In Figure 2.35 the new fracture j is placed into the simulation and checked with the previous fracture i. Firstly, the coordinates of the intersection point are calculated using a vector method. When the intersection coordinates have been found, a check is performed to see if the intersection coordinates fall between the limits of both lines. Where the intersection is not within the limits of both lines (case A) no termination has occurred. Conversely, where the intersection coordinates fall between the limits of both lines (case B) then an intersection must have occurred. When an intersection occurs, a stochastic process is employed to decide whether to terminate the fracture or not. Initially the discontinuity set number of the fracture j is looked up from the data matrix. A randomly generated number is used to determine the outcome of the termination event using the probability of termination associated with the relevant discontinuity set number. When a termination occurs, a correction procedure is invoked to retain the discarded trace length. The final action is to update the endpoint coordinates and change termination codes for the affected discontinuity.

99

Chapter 2: Rock Mass Characterisation

End

End Intersection coordinates

Intersection coordinates

j

Start

Start

Start i

j

End

Start

(A) No fracture intersection

i

End

(B) Fracture intersection

Figure 2.35: Check for fracture intersections This method is used several times so that every added discontinuity in the simulation is checked with every existing discontinuity. This uses the previously ordered data list, starting with the earliest existing fracture. When all the existing fractures have been checked, the next fracture on the list is introduced into the simulation. When all fractures have been modelled, the termination part of the program is finished and a check on the discontinuity statistics is initiated to make sure that the new rock mass simulation has an acceptable statistical fit to the original modelled data.

2.7 2.7.1

ROCK MASS CLASSIFICATION SCHEMES Introduction

Usually during the feasibility and preliminary design stages of a project, very little detailed information is available on the rock mass properties or on the stress and hydrologic regimes applying. In these cases, rock mass classification schemes may be used in an attempt to extrapolate previous experience gained in the rock mass concerned or elsewhere. These classification schemes seek to assign numerical values to those properties or features of the rock mass considered likely to influence its behaviour, and to combine these individual values into one overall rating for the rock mass. Through correlations with previous experience, rock mass classification schemes can be used to make initial estimates of support requirements and of the strength and deformation properties of the rock mass. A number of the more widely used rock mass classification schemes have been applied in engineering design analyses for caving mines. The systems that have been utilised in this way include the RMR (Bieniawski 1974, 1976), Q (Barton et al 1974) and MRMR (Laubscher 1990) rock mass classification schemes. The use of these classification systems in cave

100

Chapter 2: Rock Mass Characterisation

engineering and the rock mass characterisation data required for their application are outlined below. 2.7.2

RMR system (Bieniawski 1974, 1976)

The Geomechanics Classification or Rock Mass Rating (RMR) system introduced by Bieniawski (1973, 1974) provided, partly on a subjective basis, a measure of the quality of rock masses for the purpose of preliminary support design in tunnelling. The RMR system has since been modified a number of times by Bieniawski (1976, 1979, 1989) and extended by a number of authors ostensibly to render it more suitable for particular applications (eg Romana 1985, Laubscher 1990). To classify a rock mass using the RMR system the rock mass is separated into a number of geotechnical zones or structural units having similar rock material and discontinuity properties. The boundaries delineating these zones will most often be geological contacts, major faults, and weathering profiles. Each unit is then rated separately according to the intact material strength (uniaxial compressive strength or point load strength), the RQD (Deere 1964), the discontinuity spacing and condition, and ground water inflow. Each of these parameters is given a rating (see Table 2.6) and the summed ratings give the basic RMR of that rock mass. This basic RMR can then be modified according to discontinuity orientation with respect to the excavation. One application of the RMR system, which uses the 1976 version of the system (which will be referred to as RMR76), is in the derivation of the Hoek and Brown (1980, 1997) rock mass strength parameters, m and s. Another application of the RMR system is in the estimation of tunnel support requirements (Bieniawski 1976, Hoek and Brown 1980). Table 2.6: The 1976 version of the RMR system (Bieniawski 1976) Parameter

Rating

Uniaxial Compressive Strength (UCS)

0 to 15

Rock Quality Designation (RQD)

3 to 20

Discontinuity Spacing

5 to 30

Discontinuity Conditions

0 to 25

Groundwater

0 to 10

Basic RMR76

8 to 100

Orientation Adjustment (for tunnels)

0 to –12

The procedure for calculating the RMR for the rock mass is to assess the spacing and condition for each of the identified discontinuity sets separately. For example, if there are three

101

Chapter 2: Rock Mass Characterisation

discontinuity sets in the rock mass, three independent spacing and discontinuity condition assessments are made. The different discontinuity sets are then assessed to determine which will have the greatest affect on rock mass instability (lowest summed spacing and condition values). Only the spacing and condition values for the controlling discontinuity set are then used to derive the rock mass rating and the other discontinuities are ignored (except when incorporated into the RQD measurement). The RMR system has been successfully utilised in many practical design situations in a wide range of applications and remains a useful way of articulation between geologists, mining engineers and geotechnical engineers. The practical utilisation of the RMR classification system will be considered by reviewing the individual components of the system. Strength of intact rock material The RMR system assigns a value of between 0 and 15 points to represent the intact rock strength. This value is calculated from uniaxial compressive strength or point load index test results. It has been argued by Meyers et al (1993) that the behaviour of rock masses under low stress conditions, such as in surface outcrops, is influenced more by the shear strength properties of the discontinuities than by the compressive strength of the intact rock material. The only way the shear strength of discontinuities can be considered in the RMR system is as discontinuity roughness. Development work in mines using caving methods will rarely be in low stress conditions so this argument is invalid for most cases of cave design, including caveability and fragmentation studies. Rock Quality Designation(RQD) The RQD (Deere 1964) remains the most commonly used discontinuity characterisation parameter for drill core. The RQD is defined as the percentage of core recovered in intact pieces of 100 mm or more in length in the total length of a borehole, ie

RQD = 100 ×

∑ Core pieces > 100 mm Total length of core

(2.18)

It is normally accepted that the RQD should be determined on a core of at least 50 mm in diameter which should have been drilled with double barrel diamond drilling equipment (Hoek and Brown 1980). The use of RQD as a measure of discontinuity frequency is unreliable because • •

it is reliant on the ability of the geologist logging the core to discriminate between natural fractures and those caused during blasting or drilling; it is reliant on the shear strength of the rock material being drilled;

102

Chapter 2: Rock Mass Characterisation

•

•

good core recovery is dependent on the drilling practice. Machine condition and the experience and care exercise by the driller are almost impossible to take into account; and RQD is not a good measure of the better rock mass conditions. If a rock mass has one uniformly spaced discontinuity set with a spacing of 0.l m or 5 m, the RQD will be 100 in both cases (where a 50 mm diameter core is used).

The RQD value does not discern discontinuity orientation. If the rock mass is anisotropic, the orientation of the drilling direction to the discontinuity set will be of vital importance.

Case A - RQD 100% Spacing ≈ 125 mm Persistent joint set

Drilling direction

Case B - RQD 0% Spacing ≈ 85 mm Figure 2.36: Example highlighting the importance of drilling direction on RQD

Figure 2.36 shows an extreme example of the influence of drilling orientation on RQD. In this example, the same rock mass may be classified as very good (Case A) or as very poor (Case B) when using the guidelines proposed by Deere (1964). Some authors (eg Laubscher 1990) have suggested methods of dealing with this problem by considering only intact cylinders of rock (see Figure 2.37).

103

Chapter 2: Rock Mass Characterisation

(Deere 1964) (Deere, 1964)

(Laubscher 1990) (Laubscher, 1990)

Solid cylinder of intact rock

Figure 2.37: Original and Laubscher modified RQD However, counting only solid cylinders of intact rock is also problematic as it can also lead to orientation bias. If discontinuities are equally spaced 110 mm apart then using the traditional (centre-line) method, the RQD will always be 100%, independent of drilling direction. This is not the case if only solid cylinders of rock are counted in the RQD measure. If the drilling direction is perpendicular to the mean discontinuity direction then the RQD will again be 100%. However, if the drilling direction is at an angle of 45° to the mean discontinuity direction, the RQD will be 0% as there will only be 89 mm of solid intact core between the intersecting discontinuities (as measured during core logging). This problem becomes even worse where the core intersects discontinuities at a very flat angle because the measurement process ignores core pieces that happen to have been drilled with a small subtended angle to one discontinuity in otherwise massive rock (see Figure 2.4). Spacing of discontinuities The rating for discontinuity spacing varies from 30 to 5 and is determined using Table 2.7. Table 2.7: RMR discontinuity spacing parameter (Bieniawski 1976) Spacing of discontinuities

>3m

1-3m

0.3 - 1 m

50 - 300 mm

< 50 mm

Rating

30

25

20

10

5

The experience in using the RMR system for one project was that these groupings of discontinuity spacing were too insensitive in the cases studied (AMIRA, JKMRC and Rock Technology 1997). Many of the discontinuity sets were found to have spacings of between 0.3 and 1 m. The same rating was derived whether the discontinuity sets had a mean spacing of 0.35 m or 0.95 m. These present a theoretical difference of in situ block sizes of 1 m3 to 0.027 m3 in a simple orthogonal three discontinuity set case. Rock mass strength, caveability,

104

Chapter 2: Rock Mass Characterisation

fragmentation and support requirements would be expected to be different for a rock mass containing such different block sizes. Condition of discontinuities The discontinuity condition is given a rating from 0 to 25, depending on the descriptions given in Table 2.8. These descriptions and ratings may be used to provide estimates of discontinuity shear strength. Table 2.8: RMR discontinuity condition parameter (Bieniawski 1976) Condition of discontinuities

Rating

Very rough surfaces Not continuous No separation Hard discontinuity wall rock 25

Slightly rough surfaces Separation < 1mm Hard discontinuity wall rock

Slightly rough surfaces Separation < 1mm Soft discontinuity wall rock

20

12

Slickensided Soft gouge OR > 5mm thick OR Gouge Discontinuities 5mm Continuous Discontinuities discontinuities open 1-5mm Continuous discontinuities 6 0

General observations on the use of the RMR system The greatest advantage that the RMR system provides is its ease of use in the mining environment which aids it use as a communication tool for mining engineers, geologists and geotechnical engineers. The RMR system has provided a useful tool in conjunction with the Hoek-Brown rock mass strength criterion in the estimation of rock mass strength. While the process has worked well for rock masses with a RMR greater than 25, it does not work for very poor rock masses since the minimum value which the RMR can assume is 18 (Hoek et al 1995). 2.7.3

Q system (Barton et al 1974)

The Q system of rock mass classification was developed for tunnel support in hard rock and is based on a numerical assessment of the rock mass quality using six parameters: 1. 2. 3. 4. 5. 6.

Rock quality designation Discontinuity set number Discontinuity roughness number Discontinuity alteration number Discontinuity water reduction factor Stress reduction factor

105

RQD Jn Jr Ja Jw SRF

Chapter 2: Rock Mass Characterisation

These parameters are grouped into three quotients which are estimates of: • • •

Relative block size Inter-block shear strength Active stress

[RQD/Jn] [Jr/Ja] (≈ tanφ) [Jw/SRF]

The overall rock mass quality factor Q is equal to the product of the three quotients:

⎡ RQD ⎤ ⎡ Jr ⎤ ⎡ Jw ⎤ Q=⎢ × × ⎣ Jn ⎥⎦ ⎢⎣ Ja ⎥⎦ ⎢⎣ SRF ⎥⎦

(2.19)

Numerical values for the six original parameters are obtained from published tables. These tables and the resulting Q values, unlike the RMR system, have remained relatively unchanged over the years. Only one modification has been made to the SRF parameter to allow for rock bursting conditions (Grimstad and Barton 1993). When using the Q system for Hoek-Brown strength or yield estimation, or in the Mathews’ stability chart (Mathews et al 1980), the active stress component (the quotient of the joint water and stress reduction factors) is put equal to unity as these factors are treated explicitly in the strength or yield criterion and stress analysis. The resulting equation to calculate "Q prime" is

⎡ RQD ⎤ ⎡ Jr ⎤ Q' = ⎢ × ⎣ Jn ⎥⎦ ⎢⎣ Ja ⎥⎦

(2.20)

The individual components of the Q system will be reviewed to assess the practical applicability of the classification system. Block size quotient The approximate measure of relative block size provided by the RQD/Jn suffers from potential sampling orientation error problems because of its reliance on RQD as a measure of discontinuity intensity as discussed in Section 2.7.2. This problem can be partially overcome by using an empirical relationship between the number of discontinuities per unit volume and RQD (Palmstrom 1982): RQD ≈ 115 - 3.3 Jv

(2.21)

where Jv is the volumetric joint density or the number of discontinuities contained in 1 m3 of rock.

106

Chapter 2: Rock Mass Characterisation

The main consideration when using this equation is the correctness of the joints per unit volume estimate. The estimate is prone to sampling error as the three dimensional parameter is estimated from a one or two dimensional measurement of jointing intensity. A methodology to better estimate the discontinuity density without a sampling bias problem would increase the reliability of the method. It is possible that a discontinuity model simulation could be utilised to remove sampling biases and derive an accurate estimate of volumetric joint density. The discontinuity set number parameter (Jn) gives a rating of 0.5 to 20 depending on the number of discontinuity sets in the rock mass. Table 2.9: Q discontinuity set number classification (Barton et al 1974) Number of discontinuity sets Massive, no or few discontinuities One discontinuity set One discontinuity set plus random Two discontinuity sets Two discontinuity sets plus random Three discontinuity sets Three discontinuity sets plus random Four or more discontinuity sets, heavily fractured, "sugar cube" etc. Crushed rock, earthlike

Jn 0.5-1.0 2 3 4 6 9 12 15 20

The Jn parameter has a significant effect on the Q or Q' value. A rock mass which has many discontinuity sets forming discontinuity bounded blocks will be weaker than a rock mass with the same discontinuity intensity but not forming blocks because it has a limited number of discontinuity sets. Hence a rock mass with a greater number of discontinuity sets would be expected to cave more easily or to require more support. However, several concerns about the use of the Jn have been identified. 1.

The subjective nature of choosing the number and delineating the discontinuity sets.

2.

The determination of the number of discontinuity sets in a rock mass is prone to sampling error. If a large, well planned discontinuity mapping exercise is undertaken (with three orthogonal scanlines) more discontinuity sets are likely to be identified than in a cursory examination of just one excavation face.

3.

The concept of a random joint set is flawed. A joint is defined as a discontinuity formed to release stress within rock and along which no or very little shear movement has occurred. The type and orientation of any discontinuity is governed by the relative magnitudes of the effective stresses during propagation (Hoek 1968). The principal 107

Chapter 2: Rock Mass Characterisation

stress field within a rock mass may result from external forces such as tectonic events or the weight of overburden or from internal forces such as those resulting from contraction during cooling (Rawnsley et al 1990). 4.

Many rock masses have four or more discontinuity sets, yet where the discontinuity spacing in such sets is sufficiently large, it may be incorrect to describe the rock mass as heavily fractured or "sugarcubed".

Inter-block shear strength quotient The inter-block shear quotient [Jr/Ja] provides a good measure of the shear strength of discontinuities and can be used in estimating friction angles of the discontinuities. The very detailed treatment of discontinuity roughness and discontinuity alteration is perhaps the strongest feature of the Q-system which is not emphasised in the RMR system. The Jr and Ja parameters and the guidelines provided on their use (Loset et al 1997) are well designed and objective. 2.7.4

Modified Basic RMR or MBR system (Kendorski et al 1983)

The modified basic RMR or MBR system was developed specifically for use in establishing drift support levels in caving mines. The data from which the system was developed were collected from several block caving mines in the USA. The organisation of the MBR system is shown in Figure 2.38. It follows closely the Geomechanics Classification (RMR) system (Bieniawski 1979) and incorporates some ideas introduced by Laubscher (1981). The main differences lie in the arrangement of the initial terms and in the adjustment sequence. In the MBR system, the inputs are selected and arranged so that a rational rating is still possible using very preliminary geotechnical information obtained from drill holes. The initial ratings obtained from rock mass characterisation are shown on the left side of Figure 2.38. They include intact rock strength, RQD, discontinuity spacing, discontinuity conditions and groundwater conditions. These five parameters are the same five parameters assessed in the Geomechanics Classification (Bieniwaski 1979) and discussed in Section 2.7.2. The only difference is that the orientation parameter is not used to derive the initial MBR but is one of the mining adjustments. The MBR is an indicator of rock mass "competence", without regard to the type of opening constructed in it. The next stage is the assignment of numerical adjustments to the MBR that adapt it to the ore block development process. Input parameters relate to excavation (blasting) practice, geometry, mining depth and fracture orientation. The adjustment values are obtained from tables and charts provided by Kendorski et al (1983). The MBR is multiplied by these decimal adjustments to obtain the adjusted MBR.

108

Chapter 2: Rock Mass Characterisation

BLASTING DAMAGE AB 0.8-1.0

INDUCED STRESSES AS 0.8-1.2

FRACTURE ORIENTATION AO 0.7-1.0

MAJOR STRUCTURE S 0.7-1.1

DISTANCE TO CAVE LINE DC 0.8-1.2

BLOCK/PANEL SIZE PS 1.0-1.3

STRENGTH OF INTACT ROCK 0-15

DISCONTINUITY DENSITY (RQD, SPACING) 0-40

MODIFIED BASIC RMR (MBR) 0-100

DEVELOPMENT ADJUSTMENTS

PRODUCTION ADJUSTMENTS

ADJUSTED MBR = MBR x AB x AS x AO

FINAL MBR = AMBR x DC x PS x S

DISCONTINUTY CONDITION 0-30 PERMANENT SUPPORT CHART

ISOLATED DRIFT OR DEVELOPMENT SUPPORT CHART GROUNDWATER CONDITION 0-15

SUPPORT RECOMMENDATIONS FOR SERVICE AREAS

SUPPORT RECOMMENDATIONS FOR DRIFTS DURING DEVELOPMENT

SUPPORT RECOMMENDATIONS FOR DRIFTS DURING PRODUCTION

Figure 2.38: Organisation of the MBR system (Bieniawski 1984, after Kendorski et al 1983) 2.7.5

MRMR system (Laubscher 1990)

The mining rock mass rating (MRMR) system was first introduced by Laubscher in 1974 as a development of Bieniawki’s RMR system to cater for diverse mining situations. The fundamental difference was the recognition that in situ rock mass ratings (RMR) had to be adjusted according to the mining environment so that the final ratings (MRMR) could be used for mine design. Adjustments were introduced for weathering, mining-induced stresses, joint orientation and blasting effects (Laubscher 1990). For several years, the MRMR system has been the most widely used classification system for cave mine design. Laubscher and Jakubec (2001) have recently published some revisions to the MRMR system. Only the pre-2001 version of the MRMR will be considered here. The revised version will be discussed in Section 2.7.6. To derive the basic ‘Laubscher’ rock mass rating (RMRL90), the intact rock strength, discontinuity frequency and discontinuity condition must be assessed. Intact rock strength The first parameter in the RMRL90 classification system is intact rock strength (IRS). This IRS is defined as the unconfined compressive strength of the rock between the fractures and discontinuities. A table provided by Laubscher (1990) allocates ratings of from 1 to 20 to cater for specimen strengths from 0 to greater than 185 MPa. The upper limit was selected because 109

Chapter 2: Rock Mass Characterisation

"IRS values greater than this have little bearing on the strength of jointed rock masses" (Laubscher 1990). Discontinuity frequency parameter In the 1990 version of Laubscher’s MRMR, two techniques are proposed for the assessment of discontinuity frequency: • •

the more detailed technique is to measure the rock quality designation (RQD) and joint spacing (JS) separately, with the maximum ratings being 15 and 25 respectively; the other technique is to measure all the discontinuities and to record these as fracture frequency per metre (FF/m) with a maximum rating of 40, ie the 15 and 25 from the first method are added.

Both methodologies used to assess discontinuity intensity give a result of between 0 and 40. Laubscher states that when using the fracture frequency method, all discontinuities should be assessed. When using the more detailed joint spacing and RQD method, the joint spacing refers to the spacing of joints. Joints are defined as " an obvious feature that is continuous if its length is greater than the width of the excavation or if it abuts against another joint, i.e. joints define blocks of rock". Therefore, smaller discontinuous joints are ignored when calculating joint spacings. It is important, therefore, to have RQD measurements from drill core observations (that will take into account smaller or discontinuous jointing). It is also important to recognise the difference in the method of calculating RQD suggested by Laubscher as discussed in Section 2.7.2. The approach adopted by Laubscher (1990) in defining joints disregards some of the threedimensional characteristics of rock discontinuities. It is possible that a drive in which the mapping is carried out only just cuts (samples) the edge of a large discontinuity which may play an important role in issues such as caveability or excavation stability. This is a sampling issue that is not adequately covered in the description of the classification system. If no drill core is available to obtain RQD measurements, then the RQD needs to be estimated from excavation mapping results. It is important that this mapping takes into account the smaller discontinuities that would have been included in the RQD analysis as well as the largerscale discontinuities. Therefore, when undertaking discontinuity scanline surveys for the calculation of RMRL90 or MRMR values, orientation, trace lengths and termination of all discontinuities should be recorded. Discontinuity condition parameter The rating for discontinuity condition starts at a base of 40 and is reduced by percentage modifiers for:

110

Chapter 2: Rock Mass Characterisation

• • • • •

large scale joint expression; small scale joint expression; joint wall alteration; joint filling; and groundwater inflow.

Although it does not provide a mechanism for estimating discontinuity friction angle, it does nevertheless provide a comprehensive method to systematically account for discontinuity conditions that will have an effect on rock mass strength. It should be noted that water is accounted for in this classification system, and not as an effective stress component as in the Hoek-Brown failure criterion (Hoek and Brown 1997). Mining adjustments The sum of these ratings gives the Laubscher RMRL90. To derive the MRMR from the basic RMR L90 the adjustments summarised in Table 2.10 have to be applied. Table 2.10: MRMR mining adjustments (Laubscher 1990) Parameter

Possible adjustment %

Weathering

30-100

Orientation

63-100

Stress

60-120

Blasting

80-100

Guidelines are given for the possible adjustments identified by Laubscher (1990). However, the guidelines are ill-defined for some circumstances, which can make the adjustments highly subjective and dependent on the experience of the operator (Milne et al 1998). In particular, the adjustment for stress, which may be positive or negative, has very few guidelines. A large number of factors which may effect the stress adjustment are listed but no detailed guidelines are presented on how to derive the adjustment factors. 2.7.6

Revised MRMR system (Laubscher and Jakubec 2001)

The methods introduced by Laubscher and Jakubec (2001) for establishing the In situ Rock Mass Rating or IRMR of a given rock mass and deriving the MRMR for a particular mining application, are summarised in the flow chart shown in Figure 2.39.

111

Chapter 2: Rock Mass Characterisation

INPUT DATA IRS Mpa X 80% size adj.

JOINT SPACING Rating = 0 - 35

RBS adjustment 60 – 100%

RBS value

JOINT CONDITION Rating = 0 - 40

Adjustment for cemented joints 70 – 100%

Rating

Mpa Mpa

0 - 25

JOINT OVERALL Rating = 0 - 75

IRMR

= 0 - 100

RMS = MPa PRESENTATION COMMUNICATION BASIC DESIGN ADJUSTMENTS Weathering / Orientation / Induced Stress / Blasting / Water (30 – 100%)

(63 – 100%)

(60 – 120%)

(80 – 100%)

(70 – 110%)

MRMR 0 - 100

DRMS MPa

MAJOR STRUCTURES DETAILED DESIGNS CAVEABILITY GEOMETRY

STABILITY PILLARS

FRAGMENTATION CAVE ANGLES

SUPPORT

SEQUENCE PIT SLOPES

Figure 2.39: Flow chart for calculating IRMR, MRMR and DRMS (Laubscher and Jakubec 2001)

112

Chapter 2: Rock Mass Characterisation

The essential features of each of the steps and the main changes made from Laubscher’s previous RMRL90 and MRMR system are summarised in the following points. 1.

The intact rock strength (IRS) is the unconfined compressive strength derived from the testing of rock cores. A nomogram is given for determining the "corrected" value of the IRS for the case in which the rock contains intercalations of weaker material.

2.

To obtain the rock block strength (RBS) from the "corrected" IRS, two adjustments are made. The first is a multiplication by 0.8 to allow for the size effect when applying the results of small scale laboratory tests to the field scale. The second is an adjustment for the presence of fractures and veins which will reduce the strength of the rock block. Laubscher and Jakubec (2001) give a procedure in which an adjustment is determined from the product of the inverse of the Moh’s hardness of the fractures and veins and the fracture and vein frequency per metre. Thus RBS = IRS x 0.8 x fracture/vein adjustment. A rating of 0 – 25 for the rock block strength is then read from a graph.

3.

In the revised system, the joint spacing rating for open joints is reduced to a maximum of 35 and is determined on the basis of the joint spacings of one, two or three joint sets and no more. The ratings differ from those used previously. If there exists in the rock mass a set of cemented joints in which the cement strength is less than that of the rock material, a further downwards adjustment to the joint spacing rating of 70 to 100% is made depending on the number of cemented joint sets present (one or two) and the cemented joint spacing.

4.

The maximum joint condition (JC) rating for single joints remains at 40 but the joint condition adjustments have been revised. A chart is given for determining the JC rating for multiple joint sets having differing joint conditions.

5.

The overall joint rating of 0–75 is the sum of the joint spacing (0–35) and the joint condition (0–40) ratings.

6.

The IRMR is calculated as the sum of the rock block strength and overall joint ratings.

7.

The IRMR may be multiplied by adjustment factors for weathering (30-100%), orientation (63-100%), induced stress (60-120%), blasting (80-100%) and water (70-100%) to give the Mining Rock Mass Rating or MRMR. Laubscher and Jakubec (2001) give guidelines and tables for use in determining these adjustment factors.

8.

The following quotation from Laubscher and Jakubec (2001) is considered to be especially important in the present context: "The adjustment procedure has been described in previous papers, where it was stated that the adjustment should not exceed two classes, but, what was not made clear is that one adjustment can supersede another and that the total adjustment is not likely to be a multiplication of all the adjustments. 113

Chapter 2: Rock Mass Characterisation

For example, a bad blasting adjustment would apply in a low stress area but in a high stress area the damage from the stresses would exceed that of the blasting and the only adjustment would be the mining induced stress. The MRMR for a caveability assessment would not have blasting as an adjustment, nor would it have weathering as an adjustment unless the weathering effects were so rapid so as to exceed the rate of cave propagation as a result of the structural and stress effects. The joint orientation and mining induced stress adjustments tend to complement each other. The object of the adjustments is for the geologist, rock mechanics engineer and planning engineer to adjust the IRMR so that the MRMR is a realistic number reflecting the rock mass strength for that mining situation." 9.

The design rock mass strength (DRMS) is the RMS reduced by the same factor as that applied to the IRMR to produce the MRMR. Thus DRMS = RMS x MRMR / IRMR.

10.

Laubscher and Jakubec (2001) do not indicate whether or not the previously published correlations of various engineering behaviours with MRMR, most notably that given by Laubscher’s caving chart (see Figure 3.1 and Section 3.2) will change as a result of the changes made to the calculation of MRMR.

2.7.7

Geological Strength Index (GSI)

As part of the continuing development and practical application of the Hoek-Brown empirical rock mass strength criterion to be discussed in Section 2.8, Hoek (1994) and Hoek et al (1995) introduced a new rock mass classification scheme known as the Geological Strength Index (GSI). The GSI was developed to overcome some of the deficiencies that had been identified in more than a decade of experience in using Bieniawski’s Rock Mass Rating (RMR) with the rock mass strength criterion. This brief account of the GSI is based on that given by Marinos and Hoek (2000) and should be read in conjunction with Section 2.8. The GSI is an index developed specifically as a method of accounting for those properties of a discontinuous or jointed rock mass which influence its strength and deformability. The strength of a jointed rock mass depends on the properties of the intact pieces of rock and upon the freedom of those pieces to slide and rotate under a range of imposed stress conditions. This freedom is controlled by the geometrical shapes of the intact rock pieces as well as by the condition of the surfaces separating them. The GSI seeks to account for two features of the rock mass – its structure as represented by its blockiness and degree of interlocking, and the condition of the discontinuity surfaces. Using Figure 2.40, the GSI may be estimated from visual examination of exposures of the rock mass or borehole core. It will be noted that the GSI does not include an evaluation of the uniaxial compressive strength of the intact rock pieces and avoids the double counting of discontinuity spacing as in the RMR system.

114

Chapter 2: Rock Mass Characterisation

Figure 2.40: Geological Strength Index (GSI) for jointed rock masses (Hoek 2003)

Although the origin and petrography of the rock are not represented in Figure 2.40, the rock type will usually constrain the range of GSI values that might exist for rock masses of that rock type. Marinos and Hoek (2000) present a series of indicative charts which show the most probable ranges of GSI values for rock masses of several of generic rock types. Figure 2.41 shows one of the charts for a group of rock types encountered in some block caving operations, the ultrabasic rocks or ophiolites (mainly peridotites and diabases). Marinos and Hoek (2000) note that, even when relatively unweathered, a characteristic of these rocks is that their discontinuities may be coated by weak minerals produced by alteration or dynamic metamorphosis. This translates their locations towards the right of the GSI chart compared to fresh igneous rocks, for example. The ophiolites may be transformed into serpentinites which, as indicated in Figure 2.41, can have low GSI values and produce very weak rock masses.

115

Chapter 2: Rock Mass Characterisation

Figure 2.41: The most common GSI ranges for typical ophiolites (Hoek 2003) 2.7.8

Conclusions

Rock mass classification systems are well utilised and will likely remain an integral part of rock mass characterisation and engineering design applications in cave mining. Most of the issues raised in reviewing the rock mass classification systems involve the treatment of rock mass discontinuities in the classification systems. A number of sampling issues associated with rock mass discontinuities have been raised and the use of RQD as a measure of discontinuity intensity has been questioned. This highlights the need to develop and use more comprehensive methods of characterising the rock mass discontinuities. With a more statistically rigorous treatment of discontinuity data it may be possible to remove some of the ambiguity surrounding some discontinuity measures. It may even be possible to develop confidence limits on rock mass classification ratings (Harries 2001).

116

Chapter 2: Rock Mass Characterisation

2.8 2.8.1

THE MECHANICAL PROPERTIES OF ROCK MASSES Scope

In developing rational solutions to a range of geomechanics problems encountered in the design of block and panel caving mines, it is necessary to define the mechanical properties of the rock mass, usually represented by its stress-strain behaviour. Important aspects of this behaviour are the constants relating stresses and strains in the elastic range, the stress levels at which yield, fracturing or slip occurs within the rock mass, and the post-peak stress-strain behaviour of the fractured or "failed" rock (Brady and Brown 1993). The collection of data for use in estimating some of these properties is part of the rock mass characterisation process. In some problems, it is the behaviour of the intact rock material that is of concern. This will be the case when considering the excavation of rock by drilling and blasting, or when considering the stability of excavations in good quality, brittle rock subject to rock burst conditions (see Section 10.3). In other cases, the behaviour of single discontinuities, or of small numbers of discontinuities, may be of paramount importance. Examples of this class of problem arise in the design of extraction level excavations to be considered in Chapter 6. They include the equilibrium of blocks of rock formed by the intersections of three or more discontinuities with the roof or wall of an excavation, the intersection of one or more discontinuities of critical orientation with a brow (see Section 6.4.3), and cases in which slip on a major fault must be considered. A different class of problem is that in which the rock mass must be considered as an assembly of discrete blocks as in the example shown in Figure 1. 7. In this case, the normal and shear force-displacement relations at face-to-face and corner-to-face block contacts are of importance in the analysis. Finally, it is sometimes necessary to consider the global response of a jointed rock mass in which the discontinuity spacing is small on the scale of the problem domain. Caving of jointed rock masses and subsidence to surface are obvious examples of this class of problem. Figure 2.42 illustrates the transition from intact rock to a heavily jointed rock mass with increasing sample size in a hypothetical rock mass surrounding an underground excavation such as an extraction level drift. It is beyond the scope of this book to consider in detail the full range of problems outlined above. A useful introduction to a range of aspects of rock and rock mass strength and deformability is given by Brady and Brown (1993). Here, emphasis will be placed on the overall strength, and to a lesser extent, the deformability, of jointed rock masses. In the process, some reference will be made to the strength of intact rock.

2.8.2

The Hoek-Brown Empirical Strength Criterion

The reliable determination of the global mechanical properties of large masses of in situ discontinuous rock has long been one of the most challenging problems met in the field of rock mechanics. In an attempt to provide a "first pass" method of estimating the strength of jointed 117

Chapter 2: Rock Mass Characterisation

rock masses for use in underground excavation design, Hoek and Brown (1980) developed an empirical rock mass strength criterion based on their earlier research into the brittle fracture of rock (Hoek 1968) and the mechanical behaviour of discontinuous rock masses (Brown 1970). The criterion took the strength of the intact rock as its starting point and introduced factors to reduce the strength on the basis of the spacing and characteristics of the joints within the rock mass. Hoek and Brown (1980) used the 1976 version of Bieniawski’s Rock Mass Rating (see Section 2.7.2) as an index of the geological characteristics considered likely to influence the mechanical properties of the rock mass. Because of a lack of suitable alternatives, the Hoek-Brown empirical rock mass strength criterion was soon adopted by rock mechanics practitioners and sometimes used for purposes for which it was not originally intended and which lay outside the limits of the data and methods used in its derivation. Because of this, and as experience was acquired with its practical application, a series of changes were made and new elements introduced into the criterion. Hoek and Brown (1997) consolidated the changes made to that time into a comprehensive account of the criterion and gave a number of worked examples to illustrate its application in practice. A further update is given by Hoek et al (2002). The summary of the criterion presented here is based on those of Hoek and Brown (1997), Hoek et al (2002) and Marinos and Hoek (2000). Based on analyses of a wide range of triaxial test data on rock samples, Hoek and Brown (1980) proposed that the peak strength of the intact pieces of rock in a rock mass of a given rock type could be represented by the equation ⎛ σ1′ = σ3′ + σ ci ⎜ m i ⎜ ⎝

0.5 ⎞ σ3′ ⎟ + 1.0 ⎟ σ ci ⎠

(2.22)

where σ1′ and σ3′ are the major and minor principle effective stresses at peak strength, respectively; mi is a parameter obtained by the statistical analysis of a set of triaxial compression tests on carefully prepared 50 mm diameter core samples of the intact rock; and σci is the measured uniaxial compressive strength of the intact rock. As the result of an analysis of triaxial test results on a wide range of rock types, it has been found that preliminary or approximate values of the parameter mi can be obtained from Table 2.11. The values in parentheses in Table 2.11 are estimates only. The range of values quoted for each rock group depends on the granularity and interlocking of the crystal structure; the higher values are associated with tightly interlocked and more frictional characteristics. It should be noted, however, that the values given in Table 2.11 are indicative only and that the value of mi for a given rock is established most reliably by triaxial testing. It is important that a suitable range of values of σ3 be used in carrying out the tests used to determine mi values.

118

Chapter 2: Rock Mass Characterisation

Hoek and Brown (1980) used the range 0 < σ3 < σci in deriving the original values of σci and mi and it is recommended that this range continue to be used in practice (Hoek and Brown 1997). The generalised Hoek-Brown strength criterion for jointed rock masses is given by

σ 1′ = σ 3′ + σ ci

⎛ ⎞ σ 3′ ⎜ mb + s⎟ ⎜ ⎟ σ ci ⎝ ⎠

a

(2.23)

where mb is the reduced value of the material constant m for the jointed rock mass, and s and a are parameters which depend on the characteristics or quality of the rock mass. The values of mb and s are related to the GSI for the rock mass by the relationships ⎛ GSI - 100 ⎞ m b = m i exp⎜ ⎟ 28 ⎝ ⎠

(2.24)

⎛ GSI - 100 ⎞ s = exp ⎜ ⎟ 9 ⎝ ⎠

(2.25)

and

In the initial 1980 version of the criterion, the parameter, a, took a constant value of 0.5. Subsequently, for GSI < 25, this value was increased to a = 0.65 -

GSI 200

(2.26)

which means that for rock masses of very poor quality, a ~ 0.65. Hoek et al (2002) have introduced a new expression for a which applies over the full range of GSI values:

a = 0.5 +

(

1 -GSI/15 -20/3 e -e 6

)

(2.27)

Note that for GSI > 50, a ~ 0.5.

119

Chapter 2: Rock Mass Characterisation

Table 2.11: Values of the constant mi for intact rock by rock group (Hoek 2003) Rock type

Texture Class

Group

SEDIMENTARY

Clastic

Carbonates NonClastic

Coarse

Medium

Fine

Very fine

Conglomerates * Breccias *

Sandstones 17±4

Siltstones 4±2 Greywackes (18 ±3 )

Claystones 7±2 Shales (6 ±2) Marls (7 ±2)

Crystalline Limestone (12 ± 3)

Sparitic Limestones (10±2) Gypsum 8±2

Micritic Limestones (9±2) Anhydrite 12±2

Evaporites

Chalk 7±2

Organic

Hornfels Ouartzites (19±4) 20 ±3 Metasandstone (19 ± 3) Slightly foliated Migmatite Amphibolites 26 Gneiss (29 ±3) ±6 28 ±5 Foliated** Schists Phyllites Slates 12 ±3 (7 ± 3) 7±4 Granite Diorite 32 ±3 25 ±5 Light Granodiorite Plutonic (29 ± 3) Gabbro Dolerite Dark 27 ±3 (16 ±5) Norite 20 ±5 Hypabyssal Porphyries Diabase Peridotite (20 ±5) (I5±5) (25 ±5) Rhyolite Dacite (25 ± 5) (25 ±3) Lava Volcanic Andesite Basalt 25 ±5 (25 ±5) Pyroclastic Agglomerate (19 Breccia Tuff ±3) (19 ±5) (13 ±5) Conglomerates and breccias may present a wide range of mi values, depending on the nature of the cementing material and the degree of cementation, so they may range from values similar to sandstone to values used for fine grained sediments (even under 10). These values are for intact rock specimens tested normal to bedding or foliation. The value of mi will be significantly different if failure occurs along a weakness plane.

METAMORPHIC IGNEOUS * **

Dolomites (9 ± 3)

Non Foliated

Marble 9±3

It will be seen from the above, that in order to use the Hoek-Brown criterion to estimate the strength of a given jointed rock mass, three "properties" of the rock mass are required: •

the uniaxial compressive strength of the intact rock, σci;

120

Chapter 2: Rock Mass Characterisation

•

the value of the constant mi for the intact rock; and

•

the value of the GSI for the rock mass as discussed in Section 2.7.7.

The uniaxial compressive strength of the rock mass, σcm, is obtained by setting σ3′ to zero in Equation 2.23, giving

σcm = σci.sa

(2.28)

The tensile strength of the rock mass which represents the interlocking of the particles when they are not free to dilate, is given by setting σ1/ to zero in Equation 2.23. This produces an equation which does not readily yield a simple expression for σ3/ = σtm . However, a value of σtm may be obtained by putting σ1/ = σ3/ = σtm in Equation 2.23 on the basis that the uniaxial and biaxial tensile strengths of brittle rocks are approximately equal (Brown 1974, Hoek 1968). The resulting expression is

σ tm =

-sσ ci mb

(2.29)

Analytical solutions and numerical analyses of a range of mining geomechanics problems use Mohr-Coulomb shear strength parameters rather than principal stress strength criteria of the form of Equations 2.22 and 2.23. Because Equations 2.22 and 2.23 are non-linear, the corresponding shear strength envelopes are also non-linear. This means that equivalent MohrCoulomb shear strength parameters have to be determined for a given normal stress or effective normal stress. Methods of doing this are given by Hoek and Brown (1980, 1997) and by Hoek et al (2002). An example of the use of this approach in a limiting equilibrium analysis will be given in Section 9.4.1. It is important to recognise that the Hoek-Brown strength criterion applies only to isotropic rock masses. It should not be used when failure occurs along a particular discontinuity or small number of discontinuities. The circumstances under which the criterion is, and is not, applicable are illustrated in Figure 2.42 and discussed by Hoek and Brown (1997). The major circumstances in which the criterion is likely to be of use in the analysis and design of block caving mines, is in the design of the pillars around the extraction or production level excavations, in crown pillar design and in the analysis of a number of forms of caving to surface and surface subsidence.

121

Chapter 2: Rock Mass Characterisation

Figure 2.42: Applicability of the Hoek-Brown empirical rock mass strength criterion at different scales 2.8.3

Rock Mass Deformation Modulus

A value of the deformation (or Young’s) modulus of the rock mass is required when carrying out numerical stress analyses to determine the stresses and displacements induced in the rock mass when stresses are redistributed as a result of excavation. This becomes especially important on and around the extraction levels of block and panel caving mines where the degree of excavation is high. Clearly, like its strength, the in situ deformability if a rock mass will depend on the properties of both the intact rock and the discontinuities present within the rock mass. Experience shows that this deformability can be highly variable and difficult to measure or predict. Based on the back analysis of dam foundation deformations, Serafim and Pereira (1983) proposed a relationship between in situ modulus and Bieniawski’s RMR:

E = 10

RMR-10 40

(2.30)

where E is expressed in GPa.

122

Chapter 2: Rock Mass Characterisation

This relationship was used with the early versions of the Hoek-Brown criterion. However, it was found that for many poor quality rock masses, this method gave modulus values that were too high and so a modified relationship was proposed for values of σci of less than 100 MPa (Hoek and Brown 1997):

E=

σ ci 100

⋅ 10

GSI-10 40

(2.31)

The effect of a reduction in the uniaxial compressive strength of the intact rock below 100 MPa in poorer quality rock masses is illustrated in Figure 2.43. Although Equation 2.31 has been found to work reasonably well for those cases to which it has been applied, it should be recognised that it is only an approximate method and should always be verified by field experience. It must also be remembered that although rock mass deformability is often anisotropic (Brady and Brown 1993), Equations 2.30 and 2.31 assume isotropic behaviour.

Figure 2.43: Rock mass deformation modulus as a function of GSI and uniaxial compressive strength of the intact rock (Hoek 2003)

123

Chapter 2: Rock Mass Characterisation

2.9

IN SITU STRESSES

It is well established that the behaviour of underground excavations in rock is influenced by the pre-excavation state of stress (eg Hoek and Brown 1980). The stresses and displacements induced in the rock surrounding an excavation will depend on the initial state of stress (which may, itself, have been influenced by other nearby openings), the geometry of the excavation and the constitutive (stress-strain) behaviour of the rock mass. These induced stresses and displacements will influence the stability of the excavation, the need for reinforcement of the rock mass or filling the excavation, and the initiation and propagation of caving. The role played by in situ and induced stresses in caving mechanics has been discussed in Section 1.2.2. The influence of undercut strategy and design on the stresses induced on the extraction and undercut levels will be considered in Chapter 5. Except in particular geological environments, it is usually not possible to predict in situ states of stress using the principles of mechanics. This is because both the magnitudes and orientations of stresses are influenced by a wide range of factors including tectonic history, topography, erosion, differences in the elastic constants of the lithological units and the presence of faults and other discontinuities. It must also be remembered that stress is a tensor quantity which requires the quantification of six unknowns in order to define it fully at a point. It may not be assumed that the principal in situ stresses will be oriented horizontally and vertically although, in some circumstances, it may be both reasonable and convenient to do so. Among the first significant measurements of stresses in underground mines were those made in the iron ore mines of eastern France in the early 1950s using the then recently developed flatjack method (Tincelin 1952). However, it was the pioneering work of Hast (1958) in Sweden which demonstrated that the horizontal stresses in rock could be several times the vertical or overburden stress. This result has been confirmed by a wide range of subsequent measurements and studies as reflected in compilations of measurements made in Australia and elsewhere (eg Brown and Windsor 1990, Hoek and Brown 1980, World Stress Map Project 2003, Zoback 1992). Since the 1950s, a number of approaches have been developed for the measurement of in situ stresses. As well as the original flat jack method which suffers from a number of inherent disadvantages, most emphasis has been placed on a variety of overcoring methods and on the hydraulic fracturing method which, like the flat-jack method, makes a number of sometimes limiting assumptions (Brown and Windsor 1990). For the last two decades, the state-of-the-art method has been the CSIRO hollow inclusion stress cell developed by Worotnicki and Walton (1976). Some ingenious larger scale methods of measuring local stress fields have also been used (eg Brady et al 1976). Other approaches used to estimate in situ stresses include the resolution of earthquake focal mechanisms, the interpretation of stress conditions associated with young geological features including faults, and the back analysis of wellbore breakouts

124

Chapter 2: Rock Mass Characterisation

and excavation behaviour. Overviews of these various methods of measuring and estimating in situ stresses are given by Brown and Windsor (1990) and Amadei and Stephansson (1997). Despite more than 40 years of accumulated experience in measuring and estimating stresses at mine sites, the undertaking can still be fraught with difficulty. It has been found that stress magnitudes and orientations can vary markedly with the presence of discontinuities and with changes in rock properties. Even though these effects are often observed, it is still considered essential that every effort be made to measure (from exploration openings) or otherwise estimate the mine- or orebody-scale stresses before decisions are made about the adoption of a particular mass mining method or layout. In the absence of any other source of information, recourse can be made to the several excellent databases and stress maps available on mining district, regional and world scales (eg Amadei and Stephansson 1997, Hillis et al 1999, World Stress Map Project 2003). These stress maps usually show the directions of the principal horizontal stresses in the region of interest. When relying on the data presented in these databases and stress maps, it should be remembered that not all of the results may be assumed to be accurate, particularly at shallow depths where the strains and stress components being measured are low.

125

CHAPTER 3 CAVEABILITY ASSESSMENT

3.1

INTRODUCTION

As discussed in Chapter 1, block and panel caving methods of mining require more development before the start of production than most other methods. This means that they have comparatively high initial capital costs and are relatively inflexible. Initiating and sustaining the cave govern the early productivity and economics of the operation. The ability to predict with a reasonable degree of accuracy the undercut dimensions at which caving will initiate and propagate is fundamental to the success of the mining method in most orebodies. This issue is becoming increasingly important given the considerable interest now being shown in applying caving methods to stronger rock masses. If the caveability of the orebody is not assessed with reasonable accuracy, expensive and time consuming measures may be required subsequently to initiate or sustain caving (eg Kendrick 1979, van As and Jeffrey 2000). The mechanics of, and the factors influencing, caving were discussed in Section 1.2. The major factors likely to influence caveability are discontinuity geometry and strength, rock mass strength, orebody geometry, undercut dimensions, the stresses induced in the crown of the undercut or cave and the presence of any boundary weakening. Although the importance of these factors had been recognised for some time and several attempts had been made to codify or quantify their influences (eg Coates 1981, McMahon and Kendrick 1969, Mahtab et al 1973), it was not until the development of Laubscher’s caving chart approach in the 1980s that a method of achieving this became widely available. Although not used in some caving mines, Laubscher’s caving chart (Diering and Laubscher 1987, Laubscher 1990, 1994, 2001) is the general industry standard method of assessing caveability. Numerical modelling holds the possibility of providing a more fundamental and rigorous assessment of cave initiation and propagation than empirical methods. This approach may have advantages in cases for which current experience is lacking or not well developed. The application of a numerical modelling approach to establishing caveability is discussed later in this Chapter. The use of an extended version of another empirical approach, the Mathews stability graph method, is also discussed. 126

Chapter 3: Caveability Assessment

3.2

LAUBSCHER’S CAVING CHART

3.2.1

Overview

The challenge faced in attempting to develop empirical methods of predicting caveability is to find a means of combining measures of rock mass quality, undercut geometry and induced stresses into one simple and robust tool. As shown in Figure 3.1, Laubscher has done this by plotting the value of his Mining Rock Mass Rating, MRMR introduced in Section 2.7, against the hydraulic radius, S (area/perimeter), of the undercut which is a measure of the undercut size and shape. If the orebody is elongated in one direction and has a limited width, the question can arise as to whether the minimum dimension of the undercut can influence caveability, irrespective of the value of the hydraulic radius. Data collected as part of the study to be discussed in Section 3.3 suggest that the hydraulic radius (or shape factor) is a satisfactory predictor of stability or caveability for aspect ratios of less than about three.

100 La Verna Cavern 90 Carlsbad Cave

80

BA5 (gabbro - final caving into pit)

Stable Zone Modified Rock Mass Rating, MRMR

Transitional Zone (defined by Laubscher, 1994)

Transitional Zone

Renco Freda BI 16 Shabanie L4 Stopes L4 Stopes

70

60

50

BA5 (gabbro) BI 48 Rosh Pinah BI 31

40

BI 6

30

B1 B2

B4

Stable

Durnacoal

B3 BA5 (gabbro sill) Shangani BA5 (HB) Urad BI 16 Shabanie Caving BA5 (TKB)

Transition

Zone

Caving

Andina 2nd Panel Ten Sub-6

Not Specified

King

20

King

Cassiar Cassair

10

0 0

10

20

30

40

50

60

70

80

Hydraulic Radius, (m)

Figure 3.1: Laubscher’s caving chart (from Bartlett 1998)

127

Chapter 3: Caveability Assessment

Over a period of time, starting with the chrysotile asbestos mines of Zimbabwe, Laubscher collected data from a range of caving mines and plotted on the chart a series of points representing particular mines, blocks or panels. The points plotted could represent stable (noncaving), transitional (major collapses or partially caving) or caving conditions of the blocks or panels concerned. On the basis of this information, Laubscher drew in boundary lines dividing the chart into stable, transitional and caving zones. As more data were gathered and analysed, and the method of calculating values of the MRMR evolved, the positions of the boundary lines became subject to change. When a possible new caving operation was being studied, the MRMR would be estimated and the value of the hydraulic radius required to initiate caving read from the chart. Laubscher’s chart has found widespread use in caving operations internationally. It has been especially successful when applied to the weaker orebodies for which it was developed initially. 3.2.2

The Mining Rock Mass Rating (MRMR)

Laubscher’s Mining Rock Mass Rating (MRMR) was discussed in Sections 2.7.5 and 2.7.6. The MRMR is based on a geomechanical classification system for rock masses termed the Rock Mass Rating (RMR) which is a modification of the well-known classification system developed by Bieniawski (1974, 1976) and discussed in Section 2.7.2. It will be recalled that Bieniawski’s original RMR involves the summation of numerical measures of five factors influencing the mechanical response of a rock mass: •

strength of the intact rock material;

•

Rock Quality Designation (RQD);

•

spacing of the joints;

•

condition of the joints; and

•

groundwater conditions.

Laubscher’s basic RMR (Laubscher 1977, 1984) uses the first four of these factors but measures some of them in different ways from Bieniawski and allows for the effects of water in the joint condition term. Following a series of modifications, Laubscher’s version of RMR (eg Laubscher 1994) has now diverged considerably from Bieniawski’s original and can produce significantly different ratings for given rock masses. Laubscher (1990) developed a range of mining adjustments to his RMR to account for the effects of weathering, orientation of jointing, induced stresses and blasting as shown in Table 2.10. The adjusted value of RMR becomes the MRMR. Of these factors, only the joint orientations and the induced stresses are considered likely to influence caveability. Although limiting ranges of adjustment were specified for each factor, few guidelines were given for

128

Chapter 3: Caveability Assessment

making estimates of the appropriate adjustment factors to apply within the given ranges. The decision is left to individual engineering judgment. The effect of the adjustments can be large; the MRMR can range from 50% (the minimum recommended) to 120% of the RMR. Recently, Laubscher and Jakubec (2001) revised the methods used for estimating the MRMR and the associated adjustments (see Section 2.7.6). This revised system has not been tested in a wide range of practical applications as yet and it is not clear whether or not it will require the amendment of the boundaries of the zones of stability shown in Figure 3.1.

3.2.3

Delineation of Stability Zones

As illustrated in Figure 3.1, Laubscher’s caving chart is divided into three zones - stable, transitional and caving. A total of 29 case history data points are shown on Figure 3.1 - three stable, four transitional, 17 caving and five unspecified but assumed to be stable. Stewart and Forsyth (1995) have noted that, as with all empirical methods, a stability graph approach relies heavily on the database from which it is derived. Its predictive capability is likely to improve with the availability of more reliable data from a wider range of conditions. For example, the original Mathews’ stability graph for open stope design (Mathews et al 1980) was based on only 50 case studies. The collection of more data by Potvin et al (1989), Stewart and Forsyth (1995) and Trueman et al (2000) has increased the database to about 500 cases resulting in changes to the stability zones. It is possible that the addition of more data to Laubscher’s caving chart may lead to similar changes in the current positions of the zone boundaries. Lorig et al (1995) reported some observations of Karzulovic who proposed an additional “marginal caving region” for Laubscher’s caving chart based on experience at CODELCOChile’s El Teniente and Andina mines. There is now some suggestion that the current caving zoning may under-estimate the hydraulic radius required to ensure continuous caving in some circumstances, especially in rock masses with values of MRMR of greater that 50, although the predictions have been found to be reliable in some others. For example, Lift 1 of the Northparkes E26 block cave was predicted to cave continuously at a hydraulic radius of 25 (and an area of approximately 1 hectare) but the cave was not self-propagating (van As and Jeffrey 2000). Of course, a number of geotechnical and operational factors could have combined to produce this outcome. The suggestion also arises from the Northparkes case that, in its pre2000 form, Laubscher’s method may not be as applicable to isolated, fully constrained, stronger orebodies having small footprints as it is to larger, weaker orebodies in which adjacent mined blocks may provide some stress release. In the most recent version of the stability graph known to the writer, Laubscher (2001) has recognised the possible influence of the plan shape of the block or the orebody on caveability. When the shape is approximately rectangular with an axial ratio of more than 1.5, the boundary between the caving and transitional zones is in a similar position to that shown in Figure 3.1.

129

Chapter 3: Caveability Assessment

When the shape is more closely circular with an axial ratio up to about 1.3, a second curve is used with values of hydraulic radius at a given value of MRMR being 1.2 times those in the first case. 3.2.4

Summary

Since the 1980s, Laubscher’s caving chart has been the major method used internationally to predict caveability in block and panel caving mines. It has been particularly successful when applied to the weaker and larger orebodies for which it was first developed. However, the perception has grown in recent years that it may not always provide satisfactory results for stronger, smaller and isolated or constrained blocks or orebodies. There may be insufficient case studies available, especially for rock masses having MRMR values of more than 50, to enable the three zones of stability to be delineated with a reasonable degree of accuracy over a wide range of conditions. This is not an unusual happening when an attempt is made to extend an empirical method outside the limits of the experience for which it was first developed. A practical difficulty in the application of the method is that there may be insufficient guidelines available for the inexperienced user seeking to establish values of the adjustment factors to be applied to the RMR.

3.3 3.3.1

MATHEWS’ STABILITY GRAPH APPROACH Overview

The Mathews’ stability graph (Mathews et al 1980) is very similar in concept to Laubscher’s caving chart and predates it in the open literature by several years. Mathews et al (1980) first developed the method for open stope design for mining in hard rock at depths below 1,000 m. As was noted in Section 3.2.3, the initial stability graph was based on a relatively small amount of data. A number of authors have since collected significant amounts of new data from a variety of mining depths and for a wider range of rock mass conditions to test the wider applicability of the method and have proposed modifications and extensions (Potvin et al 1989, Stewart and Forsyth 1995, Trueman et al 2000). The modifications have related largely to the delineation of the stability zones. Potvin et al (1989) made some changes to the way in which some adjustment factors were calculated but these changes have been shown to make no appreciable difference to the predictive capability of the technique (Stewart and Forsyth 1995, Trueman et al 2000). Therefore, only the original Mathews method of determining adjustment factors will be described here. The design procedure is based upon the calculation of two factors - the stability number, N, which represents the ability of the rock mass to stand up under a given stress condition, and the shape factor or hydraulic radius, S, which accounts for the geometry of the surface. The stability number is analagous to Laubscher’s MRMR, while the shape factor is identical to the

130

Chapter 3: Caveability Assessment

hydraulic radius used in Laubscher’s caving chart. For given stope surfaces, these factors are plotted on the stability graph which is divided into zones of predicted stability and instability. Stability number The stability number, N, is defined as N= Q' .A .B .C Q' is calculated from the results of structural mapping or geotechnical core logging of the rock mass using the NGI Q classification system (Barton et al 1974) discussed in Section 2.7.3 but assuming the joint water reduction parameter and stress reduction factor to be both equal to one. The rock stress factor, A, is determined from the ratio of the unconfined compressive strength of the intact rock to the compressive stress induced at the centre-line of the stope face. The induced stress is found using an elastic stress analysis package or estimated from published stress distributions (eg Hoek and Brown 1980). Figure 3.2 provides a graph which shows the relationship between the strength to stress ratio and the rock stress factor, A. The joint orientation adjustment factor, B, is a measure of the relative difference between the dips of the stope surface and the critical joint set (see Figure 3.2). The gravity adjustment factor, C, reflects the fact that the orientation of the stope surface influences its stability and is determined from Figure 3.2. To the inexperienced user, the determination of the stability number may give an inaccurate impression of the engineering rigour of this design technique. However, it must be recognised that it is difficult to separate and evaluate empirically each of the several factors that influence the stability of stope surfaces. The adjustment factors appear to have been determined largely without reference to a database of recorded observations. Nevertheless, the detailed guidelines provided for the determination of adjustment factors represent some advance over other methods. Stability zones As shown in Figure 3.3, the original Mathews stability graph contained three zones separated by transitions - a stable zone, a potentially unstable zone and a potentially caving zone. In the modified stability graph developed by Potvin et al (1989), these three zones were reduced to a stable zone and a caved zone separated by a transition. The choice of the word “caved” to represent what is essentially an unstable zone was commented on by Stewart and Forsyth (1995) who noted that the term has a particular meaning in mining which appears not to have been adhered to in the modification proposed by Potvin et al (1989).

131

Chapter 3: Caveability Assessment

Factor B Joint orientation adjustment

Factor A Rock stress factor

Factor C Design surface orientation factor

Figure 3.2: Evaluation of adjustment factors in the Mathews stability graph method (after Mathews et al 1980)

Stewart and Forsyth (1995) updated Mathews’ stability graph proposing four zones separated by three transitions - potentially stable, potentially unstable, potential major failure and potentially caving. The potentially caving zone was determined using Laubscher’s (1990) caving chart as a guide and requires validation. Nevertheless, it was intended to represent true caving. In this sense, caving is defined as occurring when the rock mass fails and collapses until all the available void space is filled with broken rock and then continues to fail when broken rock is removed from contact with the stope surface. A major failure is defined as occurring when either greater than 30% dilution occurs or when the depth of collapse into the stope walls is greater than 50% of the smaller dimension of the opening (Stewart and Forsyth 1995). Wide transition zones between the unstable to major failure and the major failure to caving zones were assumed. Mawdesley et al (2001) give a more detailed description of the development of the Mathews stability graph method.

132

Chapter 3: Caveability Assessment

Figure 3.3: The original Mathews stability graph (after Mathews et al 1980)

3.3.2

Extension of the Method

Trueman et al (2000) collected significantly more stable, minor failure and major failure case history data enabling the Mathews method to be extended to cover a much wider range of open stope sizes and rock mass characteristics. These data were combined with existing cases to produce a database of about 500 entries. The stability numbers for all these case studies were determined using the guidelines originally proposed by Mathews et al (1980). As shown in Figure 3.4, the S-N data points were plotted on log-log axes rather than the log-linear axes normally used for the Mathews method. With the increased database, a logistical regression analysis was used to define stable, minor failure and major failure boundaries. Mawdesley et al (2001) provide a detailed account of the statistical technique used.

133

Chapter 3: Caveability Assessment

Figure 3.4: Extended Mathews stability graph for open stopes based on logistic regression (Mawdesley et al 2001)

The statistical distinction between failure and major failure was not as definitive as the boundaries between stable and failure and major failure and caving. This is hardly surprising given the rather arbitrary definition of major failure. The possibility of misclassifying the two types of failure is much greater than the possibility of misclassifying a stable or a caving data point. Accordingly, Trueman et al (2000) found that it was not possible to separate failure from major failure statistically for the extended database. Although zones of stability can be defined statistically, a number of data points apparently report to the wrong stability zones. This is to be expected given the inherent variability of rock masses, data that can be somewhat subjective and the fact that the design technique is nonrigorous. Diederichs and Kaiser (1996) proposed drawing iso-probability contours to account for the uncertainty inherent in design limits. Mawdesley (2002) and Mawdesley et al (2001) calculated iso-probability contours for the stable, failure, major failure, and combined failure and major failure cases in the extended database (see Figures 3.5 and 3.6).

134

Chapter 3: Caveability Assessment

100

Figure 3.5: Isoprobability contours for stable cases (Mawdesley et al 2001)

100

Figure 3.6: Isoprobability contours for combined failure and major failure cases (Mawdesley 2002)

135

Chapter 3: Caveability Assessment

The risks associated with using the technique for open stoping can now be quantified. For example, the current boundary separating stable from failed cases (see Figure 3.4) indicates a 60% probability of the stope surface being stable, a 40% probability of either a failure or a major failure occurring and a 0% probability that continuous caving will occur (Mawdesley et al 2001). 3.3.3

Application of Mathews’ Method to the Prediction of Caveability

Purpose There are good reasons to seek to extend the Mathews stability graph approach to the prediction of caveability. As has been noted, Mathews’ approach provides detailed guidelines for the determination of the adjustment factors used. This reduces the subjectivity and degree of personal experience involved in determining the required factors in comparison with Laubscher’s caving chart method. Despite this, it must be acknowledged that a significant degree of subjectivity will always be present in applying techniques based on rock mass classifications. Nevertheless, Mathews’ approach does appear to offer some benefits in this area. Furthermore, a very large database on stope surface stability using Mathews’ approach has now been assembled. This has the potential to aid in the rational delineation of a caving zone. The importance of a large case history database for these empirical design techniques cannot be over emphasised. Caveability data collection As part of the International Caving Study Stage I, raw data were collected from blocks and panels at the Andina, El Teniente, Salvador, Henderson and Northparkes E26 mines (Mawdesley 2002). Data collected by African Consolidated Mines staff for MRMRs of less than 40 at the Shabanie and King mines were converted to Q values using a local correlation between Q and Laubscher’s RMR or RMRL. A zero stress adjustment and the presence of a flat lying discontinuity set were assumed in order to determine adjustment factors for the weak rock masses in question. Cases in which continuous caving had not initiated were designated as major failures. A number of case histories for blocky (no continuous bedding planes), strong coal mine roofs having semi-stable spans at hydraulic radii of up to 55 m were collected as part of the study. Although they were not from block or panel caving mines, these case histories should have some relevance to caveability assessment given their large undercut dimensions (Mawdesley 2002). Delineation of a caving zone The data points for the case histories collected were plotted on the extended Mathews’ stability graph. A logistical regression analysis was used to delineate a caving zone as shown in 136

Chapter 3: Caveability Assessment

Figure 3.7 (Mawdesley 2002). The line separating major failures from continuous caving on the extended Mathews stability graph does not represent a 100% probability of caving. Insufficient data are available to permit iso-probability contours to be defined accurately. The availability of additional data from well-documented case histories would allow the uncertainty in the design limits to be quantified, giving greater confidence in the use of this technique for predicting caveability.

100

Figure 3.7: Extended Mathews stability graph based on logistic regression showing the stable and caving lines (Mawdesley 2002) The influence of in situ and mining induced stresses on the caveability of an orebody has been discussed in Chapter 1. The effects of stress were readily apparent in the back-analyses of caving case histories carried out using the extended Mathews stability graph method. For example, some cases such as the Esmeralda section at El Teniente and Northparkes E26 Lift 1 had similar unadjusted rock mass classification ratings but exhibited very different caving responses. The Esmeralda section caved continuously at a hydraulic radius of 27 m (1.1 hectares of undercut area), while Northparkes E26 Lift 1 did not cave continuously at a hydraulic radius of 44 m (3.2 hectares of undercut area). However, the in situ stresses and therefore the induced stresses in the cave back for Esmeralda were much higher than those for Northparkes E26 Lift 1. Therefore, the stability numbers (analogous to MRMR in Laubscher’s method) in the two cases were significantly different despite the fact that the rock masses were perceived to have similar strengths. The difference in the stability numbers was reflected in the different caving responses. 137

Chapter 3: Caveability Assessment

Summary A review of the Mathews stability graph method suggested that it might provide an alternative method of predicting caveability, even though no caving case histories were documented. Caving case histories were collected and analysed and a caving zone delineated for an extended version of the Mathews stability graph (Mawdesley 2002). Nevertheless, more caving and transitional caving data are required to increase confidence in the use of this technique for predicting caveability. With additional data it should be possible to quantify the current uncertainty in the design limits.

3.4

NUMERICAL MODELLING APPROACHES

As noted previously, numerical modelling holds the possibility of providing a more rigorous assessment of caveability than the empirical methods just described. Numerical methods are used widely to solve stress-deformation boundary value problems in mining geomechanics for which analytical solutions cannot be obtained. This may occur when the boundary conditions, including the problem geometry, cannot be described by simple and tractable mathematical functions, the governing partial differential equations are non-linear, the problem domain is inhomogeneous, or the constitutive equations of the rock masses concerned are non-linear or insufficiently simple mathematically (Brown 1987). Clearly, most of these conditions apply in the analysis of caving which, by definition, involves non-linear and discontinuous rock mass behaviour. Numerical approaches have the advantage over empirical methods of assessing caveability of being able to treat the complex mechanics of the problem more completely and accurately. Numerical methods are also able to allow for the presence of a number of geomechanical units and for inhomogeneous responses within the problem domain. They can model major faults explicitly and represent undercut shape more completely than does the use of the hydraulic radius, or shape factor, as in the empirical methods discussed in the preceding sections. However, as Brekke and Howard (1972) have noted, rock masses are so variable that the chance of ever finding common sets of parameters and of constitutive equations that are valid for all rock masses is quite remote. Although great advances have been made in numerical modelling capabilities in the intervening years and systematic approaches to numerical modelling of rock mechanics problems have been developed (eg Starfield and Cundall 1988), Brekke and Howard’s comments remain valid. For this reason, numerical models must be shown to accurately reproduce observed caving behaviour before being used to predict caveability in practice. Most numerical models treat the rock mass as a continuum or an equivalent continuum allowing it to be assumed that the material response may be described by the equations of the 138

Chapter 3: Caveability Assessment

theories of elasticity or plasticity. Special methods of solution are required for those rock mechanics problems involving the interaction of discrete blocks of rock in which the ratio of the block size to the size of the problem domain is such that equivalent continuum behaviour may not be assumed. The most powerful and versatile method available for simulating such discontinuum behaviour is the distinct element method developed by Cundall (1971, 2001). Because of the inherently discontinuous nature of the caving process, discontinuum or distinct element approaches are intuitively attractive for use in the assessment of caveability. They also have obvious application in the modelling of particle flow as will be discussed in Chapter 7. Even if their complexity and computationally intensive nature preclude them from use in solving industrial scale problems, these methods should provide an important aid to our understanding of the caving process as illustrated by the simple example shown in Figure 1.7. An illustration of the application of the state-of-the-art three dimensional particle flow code PFC3D (Itasca 1998a) for caveability analysis, carried out by Dr Loren Lorig of the Itasca Consulting Group Inc as part of the International Caving Study Stage I, is summarised in Section 3.6 below. But before that, the application of a less numerically intensive axisymmetric continuum model to caveability prediction will be presented. Here again, the modelling described was carried out by Dr Loren Lorig as part of the International Caving Study Stage I.

3.5 3.5.1

AXISYMMETRIC CONTINUUM MODEL Model Formulation

The conceptual caving model shown in Figure 1.8 illustrates the main behavioural regions of a propagating cave. The characteristics of each of these regions were described in Section 1.2.2. It is important to note that the boundaries between these regions are diffuse rather than sharp and that they may be in different locations in different cases. Clearly, the rock mass undergoes a gradual reduction in strength from its in situ state to its caved state. In particular, the cohesion of the in situ rock mass appears to reduce during the caving process until it reaches the caved state in which the rock mass is essentially cohesionless. This observation suggests that the rock mass can be represented as a strain-softening material in which the rock mass cohesion diminishes from an initial in situ value to zero. The numerical model of caving presented here attempts to capture many of the important features of the conceptual caving model. One important assumption in the formulation of the numerical model is that the rock mass can be treated as a continuum in the sense discussed in Section 3.4. It is further assumed that the rock mass shear strength is limited by a simple MohrCoulomb failure criterion. In the discussion presented here, the two-dimensional explicit finitedifference code FLAC (Itasca 1998b) is used to model the caving process. It is possible to extend the analysis to three-dimensions using an appropriate numerical model such as FLAC3D (Itasca 1997b). 139

Chapter 3: Caveability Assessment

The axisymmetric continuum model used to study the caving process is shown in Figure 3.8. The model considers an undercut of cylindrical shape of radius R and height ho excavated at a depth, H, below the ground surface. The initial state of stress is assumed to be lithostatic. The vertical stress is given by σv = γ z (where γ is the unit weight of the material and z is the depth), and the horizontal stress is given by σh = Ko σv (where Ko, representing the ratio σh /σv, is assumed constant). Stress boundary conditions are imposed at the undercut level. The ‘support’ pressure, p(t), is reduced monotonically to simulate the extraction of material and study the resulting extension of the plastic region that develops on top of this level (Figure 3.8b). The plastic region is then interpreted as a volume of caved material available for extraction. The material on top of the undercut area is considered to be elasto-plastic, with a stress-strain response characterised by strength softening, as shown in Figure 3.9a. Two linear MohrCoulomb yield surfaces define the ‘peak’ and ‘residual” stages, respectively (Figure 3.9b). The elastic behaviour is characterised by two elastic constants, the shear modulus, G, and the bulk modulus, K. The plastic behaviour is characterised by six plastic parameters: the peak friction angle, φ; the peak cohesion, c; the residual friction angle, φr; the residual cohesion, cr; a s parameter ε crit , which defines the rate of softening (ie the accumulated plastic strain for which the residual stage is achieved); and the peak dilation angle, Ψ, which corresponds to a nonassociated flow rule.

(a)

(b)

Figure 3.8: Schematic representation of(a) the axisymmetric model, and (b) the evolution of the undercut pressure and height

140

Chapter 3: Caveability Assessment

ε scr (a)

(b)

Figure 3.9: (a) Constitutive behaviour for the material, with σ3 assumed constant, and (b) yield conditions for intact and caved materials

In order to implement a material extraction scheme in the continuum model, the undercut area is considered to be filled with a fictitious material. The elastic parameters of this material are assumed to be a fraction, fS, of the values specified for the rest of the rock mass so that the shear and bulk modulii for the undercut region are GF = fS G and KF = fS K, respectively. As noted previously, the extraction process at the undercut level is simulated by monotonically reducing the vertical pressure, p(t), at this level. Figure 3.10a shows how the support reduction is performed in steps. (The continuous type of reduction represented in Figure 3.10b has only descriptive purposes.) It is important to note that the time variable, t, included in the diagrams does not correlate directly with the actual physical time; this variable appears naturally from the explicit time formulation used by FLAC. In order to extrapolate the results obtained from the model, the number of steps (rather than the variable t) must be associated with the extraction rate at the site. Results given by the model for specified undercut geometries and mechanical properties can be used to analyse the caveability conditions above the undercut. For example, if the evolution of the cave height with continuous reduction of the undercut support pressure is as in Curve A in Figure 3.11, then the cave is stable - ie the cave will stop even as extraction of material is continued at the undercut level. On the other hand, if the evolution of the cave height is as in Curve B in Figure 3.11, the cave is unstable - ie the cave will continue to grow as material is removed from the undercut.

141

Chapter 3: Caveability Assessment

(a)

(b)

Figure 3.10: (a) Stepwise reduction of the undercut pressure, and (b) details of the pressure evolution within a reduction step

Figure 3.11: Evolution of caving height for stable and unstable cases

142

Chapter 3: Caveability Assessment

3.5.2

Material Parameters

Using the Mohr-Coulomb model for rock, the only realistic assumption for the caved mass is zero cohesion (cr = 0), zero dilation (Ψ = 0) and a non-zero friction angle (φr ≠ 0). The material, perhaps containing some large blocks, ‘flows’ towards the drawpoints and is thereby in a state that resembles the critical state of soil, characterised by a constant apparent strength and zero volume change. Because the material is broken, the mass tensile strength is zero (σt = 0), which implies a cohesion intercept of zero. It is also likely that the apparent tangent modulus of the caved material is considerably less than that of the “intact” material. The “uncaved” material (termed ‘intact’ for brevity), although containing fractures, must have a certain cohesion (c ≠ 0), because it is observed to form stable arches that exist above voids in the caved material. The primary mode of failure of the intact material is by extension fracturing parallel to the free surface, unless it is very weak (leading to deeper shear failures). Good material models for the spalling mode of failure that occurs during extension fracturing are not presently available. However, it appears (Cundall 1995, Cundall et al 1996) that the MohrCoulomb model can approximate this mode of failure given a dilation angle, ψ, greater than the friction angle, φ (eg ψ = 60°). In this case, active yielding is confined to the surface of the material, because the dilation causes a build-up of isotropic stress in the interior elements, thereby inhibiting failure. Although the yielded region appears to be a shear band when using the Mohr-Coulomb model, the real mechanism is surface spalling not shearing. The ‘transition’ from the intact to the caved material state requires some amount of strain. This transition can be captured by making the cohesion, c, dilation, ψ, and tensile strength, σt, decrease as functions of accumulated plastic shear strain. In the absence of any experimental data, the function is taken to be linear, and the parameter that determines the gradient is the s intercept, ε crit , of the softening slope with the strain axis — ie according to Figure 3.9a, the strain necessary for the strength and dilation rate to decrease to zero. The tangent bulk and shear moduli are also assumed to decrease according to a linear relation. For a simulation in which material softening is used and the response involves shear localisation, the results will depend on the element sizes. However, it is possible to compensate for this form of mesh-dependence. Consider a displacement applied to the boundary of a body. If the strain localises inside the body, the applied displacement appears as a jump across the localised band. The thickness of the band contracts until it is equal to the minimum allowed by the grid - ie a fixed number of element widths. Thus, the strain in the band is ε = u / nΔz , where n is a fixed number, u is the displacement jump, and Δz is the element width. If the softening slope is linear (ie the change in a property value, Δp , is proportional to strain), the change in property value with displacement is Δp / Δu = s / n Δz , where s is the input softening slope. In order to obtain mesh-independent results, a scaled softening slope can be input such that s = s′Δz , where s′ is constant. In this case, Δp / Δu is independent of Δz .

143

Chapter 3: Caveability Assessment

If the softening slope is defined by the critical strain (see the previous discussion), then s ε crit ∝ 1 / Δz . For example, if the zone size is doubled, the critical strain must be halved for comparable results. For purposes of illustration, simulations were carried out using the parameters listed in Table 3.1. Table 3.1: Parameters used in axisymmetric continuum analysis Property

Value

Unit weight (γ )

25 kN/m

Comments 3

Density = 2585 kg/m g = 9.81 m/s

Young’s modulus (E)

20 GPa

Poisson’s ratio (ν)

0.22

Peak friction angle ( φ )

3

2

35°

Same as residual friction angle

Unconfined compressive strength (peak)

15 MPa

Zero residual strength

Peak dilation (ψ)

60°

Zero residual dilation (ψ

= 0o )

Critical plastic strain ( ε crit )

0.01

Vertical stress at extraction level

12.13 MPa

Horizontal stress at extraction level

18.19 MPa

fp

0.1

Extraction factor

fs

20

Ratio of intact to broken modulus

ft

12

UCS/tensile strength

s

Varied (see Figure 3.9a)

(residual ft = 0)

3.5.3

Results

Two mesh sizes were used: coarse (with 38 x 40 zones), and fine (with 76 x 80 zones). In both cases, the same physical dimensions of the grid and undercut area were used. All results show a similar pattern of deformation and failure. Consider, for example, Figure 3.12 which shows s the extent of plastic yield for the fine model with ε crit = 0.01 and 2000 equilibrium cycles per excavation step. There is a strong stress discontinuity (see Figure 3.13, which shows the final state) between the caved region and the intact region, and this discontinuity migrates upwards with continued extraction. However, at some point, the discontinuity stops growing (see the history of cave height shown in Figure 3.14), and a stable roof of intact material forms. The material under this roof is generally in its residual state (ie with zero cohesion and dilation), carrying stresses that are of the order of gravitational stresses in a pile of granular material. The material flows with each extraction step (see Figure 3.15 which shows displacement vectors after 360 excavation steps). The physical interpretation is that a steady-state solution has been obtained, with the cave stabilised at a specific height and the material below the stable roof able to flow out freely whenever extraction is performed. Because FLAC operates (in this case) in small-strain mode, the material cannot actually be moved out of the caved region, but the 144

Chapter 3: Caveability Assessment

simulation provides the evidence (ie constant pressure, zero strength, and gravitational stresses) that the caved mass is in steady-state flow. For practical application, it would be desirable to devise a way of reproducing the evolving geometry as well. JOB TITLE :

(*10^2)

FLAC (Version 3.40) 5.000

LEGEND 13-Oct-98 13:57 step 720000 -7.333E+01 1.0

Heavy rockburst (strain-burst) and immediate dynamic

75-100

deformations in massive rock

188

Chapter 4: Fragmentation Assessment

12 11 10 9 8 7 6 5 4 3 2 2

4

6

8

10

12

14

Figure 4.16: Completed tessellation after the extension of joints In the implementation of the primary fragmentation module, the block identification algorithm outlined for the in situ model is used to identify the regions fully enclosed by eligible joint traces. Here again, only eligible joints are considered to form possible block boundaries. The eligibility criterion is that the joints have to be open. A joint that has been extended is considered open. This block information is then analysed and presented as block volumes using the extrapolation procedure discussed for the in situ blocks.

The block identification algorithm also records information on the extent of internal jointing. This approach inherently respects the geometric characteristics of the set of all discontinuities. It naturally provides the ability to extend pre-existing weakness planes and fracture intact rock. The method also provides the ability to consider the extent of intact rock bridges and to make primary fragmentation estimates for any applied stress level or regime. For obvious practical reasons, field data can only be obtained to indicate the combined outcomes of primary and secondary fragmentation processes. Accordingly, the contribution made to the overall fragmentation in the primary stage cannot be isolated and measured in normal caving operations. Despite the difficulty of obtaining validation data, it is considered that the concepts and approach presented above provide a sound and rational basis for further development. For example, as knowledge of crack propagation mechanisms and modelling

189

Chapter 4: Fragmentation Assessment

capability increase, it may become possible to develop a more deterministic model of joint extension.

4.7

CONCLUSIONS

In this chapter, the factors influencing fragmentation, methods of fragmentation measurement and methods of predicting fragmentation (most notably the BCF program) have been reviewed and a new method proposed for identifying in situ blocks and predicting primary fragmentation. The review of methods of fragmentation measurement concluded that digital imaging processing systems have potential for further development for use in cave mining applications if lighting and dust conditions can be controlled.

The BCF program is the only complete or near-complete method available for predicting the fragmentation produced in caving operations. Some improvements to BCF were made as part of the International Caving Study Stage I. It is likely that some of the assumptions and empirical rules used in the program could be further improved. For example, the program does not adequately represent and treat the geometrical properties of the full discontinuity network pre-existing within the caving mass. Accordingly, a new and more fundamentally based method of identifying in situ blocks and of predicting the primary fragmentation associated with the onset of caving has been developed by Eadie (2002). This method which is known as JKFrag is based on a rigorous two-dimensional tessellation procedure. It is considered to provide a reliable starting point for the development of a new or improved method of predicting caving fragmentation. The next stage of research on this problem will be to develop a method of predicting the secondary fragmentation produced within the draw column. As outlined in Section 4.2, secondary fragmentation may be produced by a variety of mechanisms, not all of which are well understood. The BCF approach to predicting secondary fragmentation uses some assumptions and rules that are considered unlikely to be universally applicable. It is considered unlikely that numerical modelling approaches will, of themselves, be able to fully resolve the problem in a practical way, although parametric numerical studies are likely to be very useful in developing improved empirical rules for primary and secondary fragmentation assessment. Therefore, the approach suggested for the development of an improved method of predicting secondary fragmentation, is to initially study the basic mechanics of the each of the major mechanisms involved using analytical and numerical models. From these studies and parametric numerical stress analyses, a set of empirical rules for each the mechanisms will be developed. In some cases, the rules used in BCF may provide an appropriate starting point, or indeed, the only practicable option. It is considered that the development of the empirical rules will also benefit from the results of distinct element numerical simulation studies of the type discussed in Appendix C. 190

CHAPTER 5 CAVE INITIATION BY UNDERCUTTING

5.1

INTRODUCTION

A

s was indicated in Chapters 1 and 3, the caving of a block or panel is initiated by mining an undercut until its hydraulic radius reaches or exceeds a critical value. As broken ore is removed progressively from the critical undercut area, the ore above it will collapse into the void so created. Vertical propagation of the cave will then occur in response to the continued removal of broken ore through the active drawpoints. Horizontal propagation of the cave will occur as more drawpoints are brought into operation under the undercut area. Experience has shown that undercutting makes a critical contribution to the success or otherwise of block and panel caving (eg Laubscher 2000). Poor planning, design, implementation and management of the undercut can jeopardise the ultimate success, productivity and costs of an operation. In particular, care must be taken to ensure that the undercut does not impose excessive abutment stresses on the surrounding rock mass and extraction level excavations which could cause delays in production and incur excessive costs through support and reinforcement requirements and rehabilitation. In summarising the nature and importance of undercutting, Butcher (2000a) has suggested that it has three aims: • • •

to extract a void of sufficient dimensions to allow caving to occur; to achieve the required undercut dimension to initiate caving with minimum damage to the surrounding rock mass; and to advance (in time as rapidly as possible) to caving hydraulic radius, initiate caving, propagate the cave and consequently reduce undercut abutment stress.

The successful implementation of the apparently simple undercutting concept in the range of circumstances met in practice requires that careful attention be paid to several factors. The examples of the background to the development of current caving practice at a number of mines given in Chapter 1 illustrate the effects of some of these factors which include: • •

the sequence of undercut and extraction level development; the relative positions of, and distances between, the undercutting front, the front of extraction level development and the extraction front;

191

Chapter 5: Cave Initiation by Undercutting

• • • •

the starting point and direction of undercut advance; the rate of undercut advance; the height of the undercut; and the shape of the undercut in both plan and vertical section.

In this chapter, accounts will be given of • • • • •

undercutting strategies; undercut design and management including the influence of the factors listed above; undercut shape and extraction methods; the influence of undercut strategies on the stresses induced in the undercut and extraction levels; and drilling and blasting practices for the mining of the undercut.

It is important to recognise that, although undercutting is the essential means of initiating caving, additional cave inducement measures may be required in order to initiate and sustain caving, especially in the higher strength rock masses for which caving methods of mining are now being used. Data collected by Flores and Karzulovic (2002b) indicate that drill and blasted slots were used to create release surfaces and assist cave initiation and propagation in more than 50% of the current caving mines studied. There is also interest in pre-conditioning rock masses by blasting or hydraulic fracturing, for example, to improve their caveability. This is one of the major topics being studied in the International Caving Study Stage II.

5.2 5.2.1

UNDERCUTTING STRATEGIES Purpose

As the examples of current practice given in Chapter 1 and the numerical analyses to be presented in Section 5.5 show, the undercutting strategy adopted can have a significant effect on the stresses induced in, and the performance of, the extraction level installations and on cave propagation. Basically, three different undercutting strategies may be used – post-, pre- and advance undercutting. To these may be added other variants such as the Henderson method. The following accounts of these undercutting strategies and of their advantages and disadvantages are based on those of Bartlett (1998) and Butcher (2000a). 5.2 2

Post-undercutting

The post-undercutting strategy is also referred to as conventional undercutting. As illustrated in Figure 1.12, undercut drilling and blasting takes place after development of the underlying extraction level has been completed. Cones, drawbells or troughs are prepared ahead of the undercut and are ready to receive the ore blasted from the undercut level.

192

Chapter 5: Cave Initiation by Undercutting

The advantages of this system are that blocks can be brought into production more quickly than with some other methods, no separate ore handling facility is required on the undercut level, drifts on the undercut level are required only for drilling and blasting and so can be 30 m apart, and the probability of ore compacting on the undercut level is very small. The main disadvantages are that, other than in low stress environments, the rock mass between the undercut and extraction levels is subjected to high and variable stress levels and support and reinforcement must be installed well ahead of the undercut stress abutment zone. This can constrain the rate of undercut advance. Butcher (2000a) suggests that, as a general guideline, the use of a post-undercutting strategy should be assessed critically when the depth of the cave is greater than 500 m, when the caving area has a hydraulic radius of greater than 17 m, and when draw horizon extraction exceeds 50%. An example of the use of a post-undercutting sequence in the BA5 block of the Premier Diamond Mine is given in Section 1.3.3. 5.2.3

Pre-undercutting

In this approach, the undercut is mined ahead of extraction level development. In some instances, the term pre-undercutting is used to describe the case in which the undercut is completed before any extraction level development is carried out. On the other hand, preundercutting may also be considered to be a variant of the advance undercutting method to be discussed in Section 5.2.4 with the development of the extraction level lagging some distance behind the undercut. The minimum horizontal distance that the extraction level development lags behind the advancing undercut is often the separation distance between the two levels. This is sometimes referred to as “the 45 degree rule”. However, in higher stress environments, it may prove necessary to use larger lag distances than those given by this rule. Even with the 45 degree rule, it may be possible that the extraction level excavations will not be in a full stress shadow zone or may “see” some abutment stress concentration. When stresses are high, this may be sufficient to cause distress to the extraction level installations. In the Esmeralda sector of El Teniente, for example, it was found that difficulties arose when the 45 degree rule was applied and that a larger lag was required. In this case, the vertical separation between the extraction and undercut levels is 12 to 15 m but it has been found most satisfactory to maintain a horizontal separation of 22.5 m between the undercut front and extraction level completion (Jofre et al 2000). The production zone is located some 45 to 60 m behind the undercut front. Advantages of pre-undercutting are that the extraction level is developed in a de-stressed environment, the undercut can be mined independently of the extraction level, support requirements on the extraction level are generally lower than in the post-undercutting method, and the broken ore in the undercut level acts as rock fill reducing the abutment loads on the undercut face to some extent.

193

Chapter 5: Cave Initiation by Undercutting

The disadvantages of this strategy include the need for a separate ore handling facility on the undercut level, possible sequencing problems between extraction horizon development and undercutting, the need to develop drawbells from the extraction level into broken rock on the undercut level, possible high stress remnants arising from the compaction of blasted undercut ore making extraction horizon development difficult, drawpoint hangups resulting from ore compaction, and slower initial production arising from these various factors. An example of the use of pre-undercutting at El Teniente is given in Section 1.3.2 and illustrated in Figure 1.12. 5.2.4

Advance Undercutting

In the advance (sometimes called advanced) undercutting strategy, undercut drilling and blasting takes place above a partially developed extraction level. The partial development on the extraction level can consist of either extraction drifts only or extraction drifts and drawpoint drifts. Drawbells are always prepared in the de-stressed zone behind the undercut, usually adhering to the 45 degree rule. Figure 5.1 illustrates a conceptual advance undercut strategy proposed for the panel cave mining of future sectors of El Teniente, Chile (Jofre et al 2000). Butcher (2000b) notes that advance undercutting is essentially a compromise between the postand pre-undercutting strategies, in that: •

draw horizon damage is reduced because, compared with the post-undercutting strategy, the extraction ratio on this horizon is decreased;

•

the cave is brought into production more quickly than with the pre-undercut strategy, reducing the problems associated with increased development times;

•

the probability of the formation of stress-inducing remnants arising from muck pile compaction is reduced;

•

a separate level is still required for undercutting but it will require a much more limited ore handling facility than in the post-undercutting strategy; and

advance undercutting is slower than post-undercutting because of the remaining extraction level development that is required after the undercut has advanced. The fact that drawbell development must be accomplished from the extraction level into broken ore in the undercut level contributes to this. On the other hand, this method obviates the need for the timeconsuming and costly repairs to the extraction level drifts that are almost inevitably required with the post-undercutting strategy.

194

Chapter 5: Cave Initiation by Undercutting

Because of its inherent advantages, the current trend in block and panel cave design is to use an advance undercutting strategy with the disadvantages that have been outlined being reduced to tolerable levels by careful planning, support and reinforcement design and equipment selection.

Undercut Sequence – Advanced U/C Panel Caving

Figure 5.1: Advance undercut panel caving, El Teniente mine, Chile (after Jofre et al 2000)

5.2.5

The Henderson Strategy

To the three main undercutting strategies must be added the Henderson or "just in time" method described in Section 1.3.3 and illustrated in Figures 1.17 and 5.2. In this approach, the drawbells are blasted with long holes from the undercut level just ahead of the blasting of the undercut itself (see Figure 5.2). This reduces the time during which the pillars and the extraction level installations are subject to high abutment stresses and damage may occur. 195

Chapter 5: Cave Initiation by Undercutting

Laubscher (2000) suggests that this strategy does not work satisfactorily in squeezing ground conditions where hole closure occurs.

Figure 5.2: Drill layout for undercut and drawbell blast, Henderson Mine, USA (after Rech et al 2000)

5.3 5.3.1

UNDERCUT DESIGN AND MANAGEMENT Purpose

As was noted in Section 5.1, a number of factors associated with the design and implementation or management of the undercut can have important impacts on undercut and overall caving performance. The purpose of the present Section is to discuss some of these issues. Among the most important of them are the method of formation and the cross-sectional shape of the undercut (flat, narrow or inclined). This factor will be discussed separately in Section 5.4. 5.3.2

Initiation and Direction of Undercut Advance

The choice of the starting or initiation point for the undercut and the preferred direction of undercut advance can be influenced by several factors including • the shape of the orebody; • the distribution of grades within the orebody;

196

Chapter 5: Cave Initiation by Undercutting

• • • •

the in situ stress directions and magnitudes; the strength of the orebody and its spatial variation; the presence and orientations of major structural features in the orebody; and the presence of caved areas adjacent to the block or panel to be undercut.

If the orebody is long and narrow in plan, a constraint will be placed on the possible directions of undercut advance. Under these circumstances, it will generally be necessary to open the undercut to the full width of the orebody and advance it in the longitudinal direction. There may be advantages in terms of productivity in retreating the cave in two directions away from a central slot or starting point as in the front caving example outlined in Section 1.2.1. However, it is more common for orebodies that may be mined by block or panel caving methods not to be of this elongated shape and to be either more approximately equi-dimensional or large in each plan dimension. In this more general case, the other factors listed above must be considered. In the case of an approximately equi-dimensional orebody, it is common for the cave to be initiated against a slot on the boundary of the orebody and advanced diagonally across the orebody as in Northparkes E26 block cave Lift 2 (Duffield 2000). Panels are usually similarly advanced on a diagonal front across the orebody as illustrated in Figure 1.3. Alternatively, the initiation point may be near the centre of the orebody with the undercut being developed progressively outwards towards the orebody boundaries as shown in Figure 5.3 for the Palabora block cave (Calder et al 2000). Operational factors as well as the distribution of grades and of the strength and caveability of the orebody need to be considered in establishing the starting point for undercutting in such cases. If, as in the case illustrated in Figure 5.3, the orebody is elongated in one direction, the issue of the minimum dimension required to achieve selfpropagating caving referred to in Section 3.2 may arise.

Figure 5.3: Planned undercut sequence, Palabora block cave, South Africa (Calder et al 2000) 197

Chapter 5: Cave Initiation by Undercutting

The influence on cave initiation of the in situ stresses and their redistribution around the undercut and the developing cave was outlined in Section 1.2.2. For the purposes of this discussion it will be assumed that the undercut drifts and the direction of cave advance are aligned with the principal horizontal in situ stresses as illustrated in Figure 5.4. If the direction of advance is perpendicular to the direction of the major principal horizontal stress, the levels of stress in the abutment ahead of the undercut will be high and will increase as the undercut advances. As the analyses and discussion to be presented in Section 5.5 will show, this will increase the likelihood of damage to the undercut drifts and the extraction level excavations. However, this effect may be an advantage in overcoming the strength and inducing caving of stronger rock masses as in CODELCO-Chile’s Andina and El Teniente mines.

Direction of cave advance σh1 σh2 σv Figure 5.4: Direction of in situ stress relative to cave advance Any spatial variation in the strength of the orebody can be expected to have an effect on the influence of the induced stresses on cave initiation and propagation. Because caving should be easier to initiate in weaker than in stronger ore, and because the stresses induced ahead of the undercut should increase as the undercut advances, it is often argued that mining should take place from a weaker to a stronger section of the orebody (Ferguson 1979). Other things being equal, there may be some advantages in terms of rates of return to offset capital costs of starting mining in any higher grade zones of the orebody. This may, however, have undesirable effects on the commissioning and operation of the processing plant. Major structural features such as faults and shear zones can have an influence on cave initiation and propagation and on the stability of undercut and extraction level excavations (eg Ferguson 1979, Laubscher 2000). A major circumstance to be avoided is the isolation of large wedges of rock that may fall or “sit down” under the influence of gravity, inhibiting cave propagation and imposing additional dead weight loads on undercut drifts and extraction level excavations. As a general rule, it is preferable to orient the advancing undercut face as close as possible to normal to the strike of any persistent structural feature or set of features. 198

Chapter 5: Cave Initiation by Undercutting

In many large caving mines, the orebody may be extracted in a series of blocks. Where possible in these cases, new blocks should be retreated from existing blocks rather than being advanced towards them. As illustrated in Figure 5.5, this prevents the creation of potentially highly stressed pillars between the two caves which can lead to the stress-induced failure of excavations in their vicinity. Ferguson (1979) gives an example of difficulties arising from this cause at the Shabanie mine, Zimbabwe.

(a)

(b)

Figure 5.5: The initiation of caving of a block adjacent to an existing caved block for (a) the preferred mining direction, and (b) an unfavourable mining direction (after Ferguson 1979) 5.3.3

Shape of the Undercut Face

Both mining experience and a consideration of the induced stresses suggest that sharp changes to, or large irregularities in, the shape of the advancing undercut face should be avoided and that the lead or lag between adjacent sections of the overall face should be minimised. Butcher (1999) has suggested that horizontal undercut lags should be generally less than 8 m to avoid significant undercut drift damage. A circular or square undercut will produce a larger hydraulic radius than a rectangular undercut of the same plan area and so should induce caving more readily. However, a flat undercut face is difficult to achieve in practice and sharp corners, especially re-entrant corners, are to be avoided. These factors argue for the adoption of a curved face with a large radius of curvature (Ferguson 1979). A face that is convex with respect to the cave should cave more readily than one that is concave. In panel caving operations, the cave front should advance across the orebody in a straight line as each of the adjacent panels is undercut as illustrated for the case of the Henderson mine in Figure 1.3. Figure 5.6 illustrates some of these desirable and undesirable features of undercut face shape and orientation in an idealised case. It must be remembered that, in practice, it is not always possible to avoid the undesirable and adopt the desirable features. The orebody boundaries may 199

Chapter 5: Cave Initiation by Undercutting

not be of the conveniently regular shape shown in Figure 5.6, and changes of lithology such as the presence of more massive and stiffer rock (which could act as a stress concentrator and increase rock burst potential) can be expected to influence the preferred undercut shape. Furthermore, the preferred undercut orientation with respect to in situ stresses may have to be modified if major discontinuities leading to rock mass instability are present.

Figure 5.6: Idealised plan illustrating some of the desirable and undesirable features of undercut shape and orientation 5.3.4

Rate of Undercut Advance

It is not an easy matter to establish the optimum rate of undercutting in a given case. To do so usually requires that a compromise be reached between a number of competing factors: •

reaching often ambitious initial production targets in order to achieve early financial returns can result in pressure to increase rates of undercutting;

•

in the advance and post-undercutting methods, the rate of undercutting cannot exceed the rate at which drawbells and drawpoints can be formed;

•

experience of the type reported in Section 1.3 shows that in high stress environments, high rates of undercut advance can increase the levels of damage to undercut drifts, pillars and extraction level excavations and can lead to rock bursting in some cases. Reducing the rate of undercutting in these cases generally reduces the extent of damage and the incidence of rock bursts (Rojas et al 2000a);

200

Chapter 5: Cave Initiation by Undercutting

•

the rate of advance must not exceed that which can be accommodated by the caving rate of the orebody. At the present time, the rate of caving can be determined only from experience. It is a topic requiring further field and theoretical research. It is clear, however, that the caving rate is a function of the stresses induced in the cave back and of their relation to the strength of the rock mass. As a general rule, to ensure that the cave remains full of caved ore, the rate of drawing ore should not exceed rate at which bulking (the excess of caved volume over in situ volume) is produced by the “natural” caving process. If the rate of draw is too large, an air gap can develop between the back of the cave and the ore pile in the cave. Sudden or massive failure of the cave back can then lead to potentially disastrous air blasts. This topic will be discussed in Chapter 10;

•

after caving has been induced, the rate of undercutting will be influenced by the height of the ore column and the extraction level layout which, in turn, influence the rate of production;

•

in some orebodies, usually those that are weaker and fragment more finely, excessively slow rates of undercutting and of removing the blasted and caved ore can lead to compaction of the ore and difficulties in achieving uniform flow and extraction of the broken ore; and

•

irrespective of the influence of these various factors, uniform temporal and spatial rates of undercut advance should be achieved for the best results.

Data collected by Flores and Karzulovic (2002b) show that undercutting rates in current block and panel caving mines may vary from 500 to 5000 m2 per month with the mean being in the range 2000 to 2500 m2 per month. Table 5.1 shows examples of rates of undercutting that have been used successfully in a number of recent cases. Table 5.1: Examples of undercutting rates Undercutting rate (m2 per month)

Mine Kimberley Mines*

2700

De Beers Premier mine BA5 panel cave

900

De Beers Premier mine BB1E block cave

1100

Esmeralda panel cave**

3000

Northparkes E26 block cave

1600

*

The Kimberley rates are influenced by the need to minimize damage to extraction level pillars.

**

The Esmeralda rates are influenced by the need to control seismicity

201

Chapter 5: Cave Initiation by Undercutting

5.3.5

Undercut Height

In the past, it has been variously thought that a high undercut would serve to limit the stresses induced in the extraction level excavations (Ferguson 1979) or to ensure that the cave propagated and the ore flowed as intended (Laubscher 1994). Furthermore, high undercuts have been seen as a source of early ore production as in the case of the Northparkes E26 mine, Lift 1 illustrated in Figure 5.7 where two undercut sublevels were used with a total height of more than 40 m (Dawson 1995). One of the assumptions has been that in the stronger orebodies in which drawbells are widely spaced to permit the use of large equipment, coarsely fragmented material would “sit” on the major apex between drawbells and would not gravitate into the drawbells so that uneven flow and draw of the broken ore would result.

Upper undercut 30m

12m

Drilled

Lower undercut

Extraction level

18m

28m

4.2x4.2m

Figure 5.7: Northparkes E26 Lift 1 high undercut geometry (Vink 1995)

From a drilling and blasting perspective, higher undercuts are easier to break because of the increased free-face area available. The risk of not achieving full breakage is greater in narrow undercuts because of the higher confinement. In addition, any small amount of hole deviation can exacerbate the confinement problem. Experience with high undercuts, particularly in the stronger orebodies, has led to the conclusion that narrow undercuts do not have the previously assumed disadvantages (Laubscher 2000). As a consequence, high undercuts have been progressively replaced at a number of operations in recent years. Some disadvantages of high undercuts have been found to be: 202

Chapter 5: Cave Initiation by Undercutting

• •

the long holes necessarily used to mine the undercut can produce irregular cave backs resulting in poor caving and fragmentation characteristics being experienced; for the reasons referred to earlier, high undercuts lead to increased initial and overall costs;

•

the slow rates of advance achieved for high undercuts may produce some of the disadvantages outlined in Section 5.3.4; and

•

mucking of swell from a high undercut can present operational problems particularly if an advance undercut is used.

Despite the advantages now being seen for narrower undercuts, several practical considerations help define a minimum undercut height in any given case. As well as the practicalities of drilling and blasting narrow undercuts to be discussed in Section 5.4 below, the degree of primary fragmentation achieved on the initiation caving is a most important factor. If the undercut is insufficiently high with respect to the block size produced in coarsely fragmenting ore, the likelihood of the formation of "pillars" of caved ore in the undercut is increased. Jofre et al (2000) describe the evolution of undercut heights and the drill layouts used in the mechanised panel caves at El Teniente. Ferguson (1979) discusses the earlier replacement of a double undercut at the King Mine, Zimbabwe, by a single narrow undercut. Despite the general recognition of the possible value of a narrow undercut, there are circumstances in which a high undercut may be beneficial. Rech et al (2000), for example, discuss the planned layout for Henderson Mines eastern section where the ore column height will be increased to 244 m. Together with this increased column height, a wider draw point spacing and a higher undercut than those presently used will allow the productivity per draw point to be increased about threefold. Table 5.2 shows the undercut heights used at some current and recent block and panel caving operations, most notably in the various sections of the El Teniente mine. The highest undercut listed in Table 5.2 is the two-stage undercut used for Lift 1 of the Northparkes E26 mine described by Vink (1995). The undercut geometry and blasting patterns used in this case are illustrated in Figure 5.7.

203

Chapter 5: Cave Initiation by Undercutting

Table 5.2: Examples of undercut heights Mine & Sector

Undercut Height (m)

5.4 5.4.1

Teniente (Ten 4 Sur B)

16.6

Teniente (Ten 4 Sur C)

13.6

Teniente (Ten 4 Sur D)

13.6

Teniente (Ten 4 Sur D)

10.6

Teniente (Ten 4 Sur Fw D)

3.6

Teniente (Esmeralda)

3.6

Teniente (Sub 6 Experiment)

16.6

Teniente (Tte. 5 Pilares)

7.0

Teniente (I-13 Tte 3)

8.6

Teniente (I-14 Tte 3)

8.6

Teniente (HP Tte 3)

4.0

Northparkes E26 Lift 1

42

Bell Canada

6

Palabora (inclined)

4

UNDERCUT SHAPE AND EXTRACTION METHOD Introduction

In addition to the factors discussed in the previous section, the shape of the undercut (in vertical section) will have a major influence on the ease and effectiveness of its formation and on its effectiveness in initiating caving. There is an especially important relationship between the design shape of an undercut and the drilling and blasting practices used for its formation. Drilling and blasting for undercutting will be discussed in Section 5.6. Traditionally, undercuts have been designed with flat or undulating roofs and flat floors broken significantly by the drawbells. The tops of the pillars or major apices left between the drawbells were then rectangular and flat. To obtain better ore flow, at least one pair of sides of the drawbells are inclined downwards, reducing the extent of the floor left at the undercut level. The major apices may also be shaped to assist ore flow and to prevent ore from stacking on the tops of the apices which may inhibit effective blasting, the flow of ore and cave propagation. Two basic methods of forming traditional “flat” backed undercuts will be discussed here, the fan and the flat undercut.

204

Chapter 5: Cave Initiation by Undercutting

An alternative to shaping only the major apex is to incline both the roof and floor to form a narrow chevron shaped undercut. This type of undercut is a relatively recent innovation advocated by Laubscher (2000). To ensure adequate flow of the broken ore to clean the floor of the undercut, the angle of undercut inclination should be greater than the angle of friction generated between the broken ore and the in situ rock. Some advantages and disadvantages of this method and some examples of designs of this type are given in Section 5.4.4. The accounts of the fan, flat and narrow inclined undercutting methods given below are based on those of Butcher (2000a). 5.4.2

Fan Undercut

Fans have probably been the most common form of drill pattern used for undercutting in the history of block and panel cave mining. They have been used with grizzly, slusher drift and LHD caves, and have been implemented with and without the development of separate undercut levels. They are relatively flexible in that they can be drilled with a range of drilling equipment, their heights can be increased to produce additional undercut tonnage and they can be adapted easily if a mine changes its mining method from open stoping or sublevel caving to block caving. The backs of the undercuts produced by fan drilling may be undulating rather than flat. This may reduce the confinement in the areas of the back that are convex downwards and help induce caving. On the other hand, there is the potential for large, relatively unbroken masses of rock to be released from these areas causing stacking or cleaning problems. The major problems with fan undercuts arise from the formation of pillars due to blast hole loss and choke blasting conditions arising from inadequate cleaning of the undercut or the stacking of blasted ore on the major apex. These problems can be overcome with good design and blasting practice as discussed in Section 5.6. A traditional fan undercut producing an undulating back at the Bell mine, Canada, is shown in Figure 5.8. Fans are also shown in Figure 5.7 illustrating the two level undercut used at the Northparkes E26 mine, Lift 1.

205

Chapter 5: Cave Initiation by Undercutting

Figure 5.8: Fan undercut used with a grizzly system at the Bell mine, Canada (Lacasse and Legast 1981)

5.4.3

Flat Undercut

As the term implies, a flat undercut is formed using flat lying drill holes rather than fans or steeply inclined holes. As a result, the undercut is narrow with a height not much greater than that of the drill drifts. A number of drilling patterns may be used in forming flat undercuts. Figures 5.9 and 5.10 show two different patterns used on the Esmeralda section of El Teniente. In the original half pillar or “John Wayne” method shown in Figure 5.9, holes were drilled from two adjacent drill drifts to meet in the middle. In the “full drilled pillar” design shown in Figure 5.9, the flat holes are drilled from one drill drift through to the adjacent drift. In either case, the best results are obtained if the holes are inclined rather than normal to the drift axes. In cases such as that illustrated in Figure 5.9, there is a danger that remnant pillars may result from poor blast hole toe breakage. This problem can be overcome by overlapping the ends of the holes in a chevron pattern. Butcher (2000a) suggests that flat or narrow undercuts are often used at deep levels because • • •

they produce higher advance rates because less drilling and charging are required; undercut blast hole loss is less because of the fact that fewer holes are required; and lower undercut heights reduce the magnitudes of the induced stresses which may otherwise cause problems.

206

Chapter 5: Cave Initiation by Undercutting

Drilled

Undercut

15m

4 x4 m 16m

Extraction Level

Drift

Drift

Drift

Drift

Figure 5.9: The original half pillar narrow flat undercut (“John Wayne”) at Esmeralda

207

Chapter 5: Cave Initiation by Undercutting

Drilled

Undercut

15m

4 x4 m 16m

Extraction Level

HW

Drift

Drift

Drift

Drift

Drift

Drift

FW

Figure 5.10: The “full-drilled pillar” undercut design at Esmeralda It is necessary that there be adequate cleaning of flat, narrow undercuts and that the stacking of blasted ore on the tops of the major apices (see Figure 5.11) be avoided. The stacking of ore during undercut formation can lead to choke blasting and the generation of remnant pillars. As shown in Figure 5.11a for the case of an advance or pre-undercut, when the swell is drawn from 208

Chapter 5: Cave Initiation by Undercutting

a flat undercut, loose blocks in the undercut back may rest down on the broken ore in the undercut. When the drawbell is developed and the drawing of the broken ore begins, the blasted ore and large blocks sitting on the major apex will not be drawn (Figure 5.11b) and a short-span stable arch may form in the cave back. Obviously, experience and high skill levels are required to satisfactorily implement flat, narrow undercuts. Flat undercuts produce lower tonnages than fan undercuts and so cannot be used as a method of ensuring high initial production rates.

(a)

(b) Figure 5.11: The ore stacking problem with a flat undercut (a) before, and (b) after drawbell development (Russell 2000)

209

Chapter 5: Cave Initiation by Undercutting

5.4.4

Narrow Inclined Undercut

Narrow inclined undercuts are now being used or considered with a view to shaping the major apex so that the undercut will be self-cleaning and stacking and choke blasting will be avoided. The chevron shape of the undercut back also enhances instability and caving of the back as an extreme of the undulating back referred to in the account of fan undercutting. A corresponding disadvantage of the method is the increased likelihood that the initial fragmentation will be coarse. Narrow inclined undercuts are being used particularly where cave mining is taking place in deep, high stress environments as at Palabora (Calder et al 2000) and for Lift 2 of the Northparkes E26 mine (Duffield 2000). As indicated in Section 5.4.1, in order for the undercut to be self-cleaning, the inclination must exceed the angle of friction generated between the blasted and the in situ rock. Experience in this and other forms of mining suggests that this angle should be greater than 45o, usually 50 - 55o. There are several possible variants of the detailed design of narrow inclined undercuts. The designs for Palabora and Northparkes E26 Lift 2 use two undercut drill drifts per drawpoint which produce flat sections of the undercut above each drawpoint. Figure 5.12 shows the narrow inclined advance undercut proposed for Palabora. Figure 5.13 is a generalised representation of this type of design illustrating some of the problems that can occur (Butcher 2000a). A particular problem illustrated is the potential to leave a remnant pillar at the top of the undercut as a result of poorly controlled drilling and blasting.

Figure 5.12: Narrow inclined undercut design for the Palabora underground mine South Africa (Calder et al 2000)

210

Chapter 5: Cave Initiation by Undercutting

Figure 5.13: Potential problem areas in a narrow inclined undercut (Butcher 2000a) Laubscher (2000) discusses two forms of an alternative an alternative design which uses only one undercut drill drift per drawpoint as illustrated in Figure 5.14. Laubscher (2000) argues that this design has advantages over the two drift design in that it provides assurance that the undercut has broken to the top of the major apex and that less development is required on the undercut level reducing costs and the induced abutment stresses. The major apex will also be higher than in the two drift design, providing more space for a secondary drilling level if required. In the first design shown in Figure 5.14a, the potential problem of leaving a remnant pillar at the top of the undercut as illustrated in Figure 5.13, again arises. This problem is addressed in the design shown in Figure 5.14a by over drilling from the advancing side. Even in this case, careful control of drilling is required. A more expensive solution is to develop a drift over the top of the major apex as illustrated in Figure 5.14b. This provides a check on the drilling accuracy and also serves as an anti-socket drift.

211

Chapter 5: Cave Initiation by Undercutting

(a)

(b) Figure 5.14: Narrow inclined undercut with a single undercut drift, (a) without, and (b) with, an anti-socket drift above the major apex (after Laubscher 2000)

5.5 5.5.1

STRESSES INDUCED IN THE UNDERCUT AND EXTRACTION LEVELS Introduction

The stability of the extraction level excavations and, to a lesser extent, that of the undercut, is critical to the efficient extraction of ore from caving mines. Observations and measurements indicate that the form and timing of the undercut has a significant influence on the stability of the extraction level drifts, primarily because of the high abutment stresses induced in the vicinity of an advancing undercut front (Bartlett 1998, Bartlett and Croll 2000, Butcher 1999, Laubscher 1994). 212

Chapter 5: Cave Initiation by Undercutting

Several factors discussed previously have the potential to influence the levels of stress induced in the extraction level excavations including the timing of the undercut relative to the extraction level development, undercut face shape, separation distance between the undercut and extraction levels, cave hydraulic radius, undercut direction and the in situ stress regime. The purpose of the analyses presented in this section is to quantify the effects of these factors on undercut and extraction level stresses. Knowledge of the likely levels of induced stress, combined with an estimate of the rock mass strength, will allow predictions to be made of the resulting levels of damage. On this basis, the undercut strategy, extraction layout design, and support design may then be optimised. Based on experience gained in cave mining operations, some undercut design guidelines have been established that are aimed at minimising stress-induced damage to extraction level excavations (Bartlett 1998, Bartlett and Croll 2000, Brumleve and Maier 1981, Butcher 1999, Lacasse and Legast 1981, Laubscher 1994). As was noted in Section 1.3.3, Bartlett and Croll (2000) describe how changing from a post- to an advance undercut resulted in reduced support requirements and a marked reduction in the rehabilitation of drifts in the BA5 panel at the Premier mine, South Africa. They also found that as the area of the undercut increased the stresses on both the undercut and extraction levels also increased. However, when continuous caving was initiated, the stress levels were observed to drop. It was also found that keeping the leads and lags between adjacent drifts to less than 8 m could reduce the damage to drifts. From observations made at the Bell mine in Canada, Lacasse and Legast (1981) note that the speed of retreat of the undercut is an important factor and that weak zones should be the starting point of an undercut where possible. As noted above, Rech et al (2000) describe the technique used at Henderson mine of developing drawbells from the undercut level close to the cave front. In this way these excavations are subjected to high abutment stresses for less time. From earlier stress measurements and observations made at the Henderson mine, Brumleve and Maier (1981) concluded that heavier support of drifts was necessary in poorer ground which had been exposed to abutment stresses for longer periods of time. This study is discussed in more detail in Chapter 8. Table 5.3 summarises five experiential design guidelines for undercutting developed by Butcher (1999). The ultimate objective of the guidelines is to reduce the level of, or minimise exposure to, high stresses in the vicinity of the undercut front. If this can be achieved, damage to the extraction or production level will be reduced.

213

Chapter 5: Cave Initiation by Undercutting

Table 5.3: Experiential design guidelines for minimising damage to the extraction level in caving mines (after Butcher 1999)

1

Guideline Use advance undercutting. If advance undercutting is not possible, minimise the percentage extracted for drift and drawpoint development in the production level.

ƒ

ƒ

2

Minimise the creation of horizontal irregularities in the undercut front.

ƒ

3

Prior to continuous caving being achieved, keep the rate of undercutting greater than the rate of damage to the extraction level.

ƒ

4

Place the undercut as high as practically possible above the production level.

ƒ

5

Advance the cave from the weakest ground to the strongest ground to achieve continuous caving as early as possible.

ƒ

ƒ

Reasons High stresses exist below the undercut front that can cause damage to pre-existing excavations in the production level. Higher extraction percentages in the production level will increase stress levels there further. Stresses concentrate in these irregularities and increase the level of damage experienced in the production level. The longer excavations are subjected to the high stresses below the undercut front, the greater the damage will be. Stresses decrease with distance below the undercut front. Stresses at the undercut front increase with the hydraulic radius necessary to achieve continuous caving. Stresses at the undercut front reduce once continuous caving is achieved.

A number of authors have used numerical models to study the stresses induced in undercut and extraction level drifts (eg Barla and Boshkov 1968, Chen 1996, Diering and Stacey 1987, Esterhuizen 1987, Flores 1993, Song 1989). Although these studies give important results, they do not allow adequately for caved block geometry or for a range of in situ stress regimes. A parametric numerical study was therefore carried out by Trueman et al (2002) as part of the International Caving Study Stage I to examine the influence of the following factors on the stresses induced in the undercut and extraction levels in a typical caving mine: • • • • •

undercut sequence; in situ stress regime; separation distance between the undercut and extraction levels; hydraulic radius of the cave; and depth below the ground surface.

214

Chapter 5: Cave Initiation by Undercutting

5.5.2

Modelling Strategy

In the study carried out by Trueman et al (2002), numerical models were used to obtain predictions of the elastic stresses on the boundaries of the excavations on the undercut and extraction levels. The finite difference code FLAC3D (Itasca 1997b) was used to analyse a three dimensional model able to represent accurately the shapes of the cave and the extraction level excavations. Because of the physical size of a block cave, a two-stage approach to stress modelling was used. In the first stage, a large-scale model of the cave itself was used to determine induced stress levels in the vicinity of the undercut and extraction level drifts. The extraction and undercut level excavations were not included in this large-scale model but the volume of rock between the extraction and undercut levels was given a lower stiffness to account for the increased extraction there. The induced stresses from the large-scale model were then transferred to small-scale models of the undercut and extraction level drifts to obtain the maximum tangential stresses in the undercut, extraction and drawpoint drift roofs. While the stresses may be higher in other areas of the drifts (eg at drift intersections), the changes in the maximum stress in the drift roofs were considered to be representative and useful for illustrating the influence of the various factors on extraction and undercut level stresses. The change in stress around the drawbells was not examined as it requires the use of a more sophisticated method for transferring stresses from the large-scale to the small-scale model (to account for the gradient in stress that occurs below the undercut). The cases examined for post- and advance undercut sequences are summarised in Table 5.4. The directions of the in situ stresses were assumed to be as shown in Figure 5.4. A preundercut may be regarded as a special case of an advance undercut and so was not included as a separate case in this study. The extraction level layout assumed was that used at El Teniente 4 South (Flores 1993). In all cases, the undercut height was constant at 4 m. In order to eliminate mining depth as a variable, all stresses were normalised to the vertical stress. Tangential stresses in the roofs of excavations were computed at a range of points from 45 m in advance of the cave front to 45 m underneath the cave. The floor to floor separation between the undercut and extraction levels was 15 m in all cases for which results are presented. A number of runs were carried out at separations ranging from 10 m to 20 m. An approximately 10% difference in the boundary stresses was noted for every 5 m difference in separation. This suggests that the boundary stresses at a 20 m separation would be about 10% lower than those presented here and those at a 10 m separation would be 10% higher.

215

Chapter 5: Cave Initiation by Undercutting

Table 5.4: Cases examined in a parametric study of extraction and undercut level stresses (Trueman et al 2002) In Situ Stress Ratio

Cave Geometry

σv:σh1:σh2 (σh1 is parallel and σh2 is perpendicular to the direction of cave advance, see Figure 5.4) 1:1:1

Length x

Hydraulic

Height (m)

Width (m)

radius (m)

60 x 60

15

150

100 x 100

25

0 75 150

1:2:1

1:1:2

1:3:2

1:2:3

200 x 200

50

150

60 x 60

15

150

100 x 100

25

150

200 x 200

50

150

60 x 60

15

150

100 x 100

25

150

200 x 200

50

150

60 x 60

15

150

100 x 100

25

150

200 x 200

50

150

60 x 60

15

150

100 x 100

25

150

200 x 200

50

150

The effect of caving height on abutment stresses was investigated by comparing the stresses at the undercut front for cave heights of 0, 75 and 150 m. At a hydraulic radius of 25 m in a hydrostatic in situ stress field (σv=σh1=σh2), the maximum induced stress was very similar for cave heights of 75 m and 150 m. However, the maximum induced stress in both these cases was 15% lower than that predicted for the case in which no caving has occurred (a cave height of 0 meters). This agrees with experiential guidelines which suggest that cave abutment stresses will drop when the cave height increases at the onset of continuous caving. The modelling strategy was partially validated by comparing the modelled induced stresses with measurements made by Flores (1993) at El Teniente 4 South. Reasonable correlations were found.

216

Chapter 5: Cave Initiation by Undercutting

5.5.3

Extraction Level Stresses – Post-undercut Sequence

Small-scale models of an LHD extraction level were used to obtain estimates of the maximum induced stress around the production drifts for an undercut to extraction level separation of 15 m. Stresses from the large-scale models were used to specify initial conditions in the smallscale models. Figure 5.15 presents the calculated maximum tangential stresses normalised to the in situ vertical stress in drift roofs for each in situ stress regime studied plotted as a function of distance from the cave boundary for varying hydraulic radii. Figure 5.15 shows that the stresses in the extraction level production drifts generally increased with hydraulic radius. This is in accord with the experiential guidelines. In general, the maximum tangential stress in the roofs of production drifts increased by about 20% with a doubling of hydraulic radius to achieve continuous caving. Exceptions were for σv: σh1: σh2 = 1:2:3 and 1:1:2 (σh1 parallel and σh2 perpendicular to the direction of cave advance) where stresses in the production drift roof were high but largely unaffected by the size of the cave. For all of the in situ stress regimes modelled except those noted above, there was a significant fall in induced stress from the peak as the cave passed over the drifts. For the majority of in situ stresses, significant stress changes were therefore apparent, with the induced stress levels rising quite significantly as the cave approached that section of the drift and falling sharply as the cave passed over the top. 5.5.4

Extraction Level Stresses – Advance Undercut Sequence

An advance undercut sequence was examined for the in situ stress cases summarised in Table 5.4. Advance undercutting was examined by making two changes to the large-scale postundercut model: • •

The undercut was extended out from the cave front by a specified distance. The broken rock in the undercut was assumed to have the same properties as the caved material. The stiffness of the extraction level ahead of the cave front was increased to represent partial development. For this study, partial development ahead of the cave front consisted of production drifts only or production and drawpoint drifts.

217

Chapter 5: Cave Initiation by Undercutting

Post undercut/σv:σh1:σh2=1:2:1 Production drift roof Tangential stress (σ θ )

3

3

2 hr= 50 m hr= 25 m hr= 15 m

In situ vertical stress (σ v )

Tangential stress (σ θ )

In situ vertical stress (σ v)

4

1

1 hr= 50 m hr= 25 m hr= 15 m

-30

-15

0

15

30

Distance from cave boundary, x(m)

45

-60

60

Post undercut/σv:σh1:σh2=1:3:2 Production drift roof

-45

Tangential stress (σ θ )

3 hr= 50 m hr= 25 m hr= 15 m

2

-15

0

15

30

45

60

Post undercut/σv:σh1:σh2=1:1:2 Production drift roof

3

4

-30

Distance from cave boundary, x(m)

In situ vertical stress (σ v )

-45

5

Tangential stress (σ θ )

2

0 -60

In situ vertical stress (σ v)

Post undercut/σv:σh1:σh2=1:1:1 Production drift roof

2

1 hr= 50 m hr= 25 m hr= 15 m

0

-60

-45

-30

-15

0

15

30

45

60

-60

Distance from cave boundary, x(m)

Tangential stress (σ θ )

-30

-15

0

15

30

Distance from cave boundary, x(m)

45

60

Post undercut/σv:σh1:σh2=1:2:3 Production drift roof

5

In situ vertical stress (σ v )

-45

4

3 hr= 50 m hr= 25 m hr= 15 m

2 -60

-45

-30

-15

0

15

30

45

60

Distance from cave boundary, x(m)

Figure 5.15: Predicted normalised tangential stresses in production drift roofs for a post-undercut (Trueman et al 2002) The small-scale model with only the production drifts or the production drifts and drawpoint drifts excavated was used to examine the stresses in a partially developed extraction level below the advance undercut. The results are shown in Figures 5.16, 5.17 and 5.18. In general, the stresses induced at the excavation boundaries again increase with hydraulic radius. For most in situ stresses and cave orientations, a doubling of the hydraulic radius leads to an approximately 20% increase in induced stress. Exceptions can occur in the roofs of drifts where the major principal stress is horizontal and parallel to the direction of drift advance. 218

Chapter 5: Cave Initiation by Undercutting

Induced stresses in production drifts are similar whether drawpoint drifts are excavated in advance of mining or not. In most in situ stress regimes, with the exception of some in which the major principal stress is horizontal and perpendicular to the drift direction, significant stress changes occur. In the main, induced stress levels are significantly lower with the cave front 15 m in advance of the section of drift being investigated. This result gives some credence to “the 45 degree rule”. However, induced stress levels continue to fall under most stress conditions, albeit at a reduced rate, as the distance between the cave front and drift section increases. The extent of the stress changes decreases significantly for excavations that are not in advance of the cave front.

Tangential stress (σ θ )

3

2 hr= 50 m hr= 25 m hr= 15 m

1 -45

-30

-15

0

15

30

Distance from undercut front, x(m)

45

1 hr= 50 m hr= 25 m hr= 15 m

60

-60

Advanced undercut without drawpoint drifts/σv:σh1:σh2=1:3:2 Production drift roof Tangential stress (σ θ )

5

4

3 hr= 50 m hr= 25 m hr= 15 m

2

-45

-30

-15

0

15

30

Distance from undercut front, x(m)

45

60

Advanced undercut without drawpoint drifts/σv:σh1:σh2=1:1:2 Production drift roof

4

3 hr= 50 m hr= 25 m hr= 15 m

2 -60

-45

-30

-15

0

15

30

Distance from undercut front, x(m)

Tangential stress (σ θ )

7

In situ vertical stress (σ v )

Tangential stress (σ θ )

In situ vertical stress (σ v)

2

0 -60

5

In situ vertical stress (σ v)

3

Advanced undercut without drawpoint drifts/σv:σh1:σh2=1:1:1 Production drift roof

In situ vertical stress (σ v)

Tangential stress (σ θ )

In situ vertical stress (σ v)

4

Advanced undercut without drawpoint drifts/σv:σh1:σh2=1:2:1 Production drift roof

45

60

-60

-45

-30

-15

0

15

30

Distance from undercut front, x(m)

45

Advanced undercut without drawpoint drifts/σv:σh1:σh2=1:2:3 Production drift roof

6

5 hr= 50 m hr= 25 m hr= 15 m

4 -60

-45

-30

-15

0

15

30

45

60

Distance from undercut front, x(m)

Figure 5.16: Predicted normalised tangential stresses in production drift roofs for an advance undercut without drawpoint drifts (Trueman et al 2002)

219

60

Chapter 5: Cave Initiation by Undercutting

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:2:1 Production drift roof Tangential stress (σ θ )

3

2 hr= 50 m hr= 25 m hr= 15 m

In situ vertical stress (σ v )

Tangential stress (σ θ )

In situ vertical stress (σ v)

4

-30

-15

0

15

30

Distance from undercut front, x(m)

45

hr= 50 m hr= 25 m hr= 15 m

-60

-45

-30

-15

0

15

30

45

60

Distance from undercut front, x(m)

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:1:2 Production drift roof

Tangential stress (σ θ )

3

hr= 50 m hr= 25 m hr= 15 m

1

In situ vertical stress (σ v)

5

4

2

1

60

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:3:2 Production drift roof

4

3 hr= 50 m hr= 25 m hr= 15 m

2 -60

-45

-30

-15

0

15

30

45

-60

60

-45

Distance from undercut front, x(m)

Tangential stress (σ θ )

7

In situ vertical stress (σ v )

Tangential stress (σ θ )

In situ vertical stress (σ v)

5

-45

2

0

1 -60

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:1:1 Production drift roof

3

-30

-15

0

15

30

45

Distance from undercut front, x(m)

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:2:3 Production drift roof

6

5 hr= 50 m hr= 25 m hr= 15 m

4 -60

-45

-30

-15

0

15

30

45

60

Distance from undercut front, x(m)

Figure 5.17: Predicted normalised tangential stresses in production drift roofs for an advance undercut with drawpoint drifts (Trueman et al 2002)

220

60

Chapter 5: Cave Initiation by Undercutting

Tangential stress (σ θ )

5 4 3

hr= 50 m hr= 25 m hr= 15 m

2 -45

-30

-15

0

15

30

Distance from undercut front, x(m)

45

2 1

60

hr= 50 m hr= 25 m hr= 15 m

-60

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:3:2 Drawpoint drift roof Tangential stress (σ θ )

7 6 5

hr= 50 m hr= 25 m hr= 15 m

4

-45

-30

-15

0

15

30

Distance from undercut front, x(m)

45

60

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:1:2 Drawpoint drift roof

4

In situ vertical stress (σ v )

Tangential stress (σ θ )

In situ vertical stress (σ v)

3

0 -60

8

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:1:1 Drawpoint drift roof

4

In situ vertical stress (σ v)

Tangential stress (σ θ )

In situ vertical stress (σ v )

6

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:2:1 Drawpoint drift roof

3

2 hr= 50 m hr= 25 m hr= 15 m

1 -60

-45

-30

-15

0

15

30

45

-60

60

-45

Distance from undercut front, x(m)

Tangential stress (σ θ )

In situ vertical stress (σ v )

6

-30

-15

0

15

30

45

60

Distance from undercut front, x(m)

Advanced undercut with drawpoint drifts/σv:σh1:σh2=1:2:3 Drawpoint drift roof

5

4 hr= 50 m hr= 25 m hr= 15 m

3 -60

-45

-30

-15

0

15

30

45

60

Distance from undercut front, x(m)

Figure 5.18: Predicted normalised tangential stresses in drawpoint drift roofs for an advance undercut with drawpoint drifts (Trueman et al 2002)

5.5.5

Undercut Level Stresses

A small-scale model assuming 4 m wide and 4 m high drifts was constructed for the undercut level and stresses obtained from large-scale models at this level imposed upon the model for the in situ stress regimes noted previously. The maximum tangential stresses induced in the roofs of drifts are shown in Figure 5.19.

221

Chapter 5: Cave Initiation by Undercutting

σv:σh1:σh2=1:2:1 Undercut tunnel roof

6

Tangential stress (σ θ )

7 6 5 4 3 hr= 50 m hr= 25 m hr= 15 m

2

In situ vertical stress (σ v)

Tangential stress (σ θ )

In situ vertical stress (σ v)

8

5 4 3

1 0

15

30

45

60

Distance from cave boundary, x(m)

0

σv:σh1:σh2=1:3:2 Undercut tunnel roof

15

Tangential stress (σ θ )

6 5 4 hr= 50 m hr= 25 m hr= 15 m

3

45

2

60

σv:σh1:σh2=1:1:2 Undercut tunnel roof

6

7

30

Distance from cave boundary, x(m)

In situ vertical stress (σ v)

8

Tangential stress (σ θ )

hr= 50 m hr= 25 m hr= 15 m

2

1

In situ vertical stress (σ v)

σv:σh1:σh2=1:1:1 Undercut tunnel roof

5 4 3 2

hr= 50 m hr= 25 m hr= 15 m

1 0

15

30

45

Distance from cave boundary, x(m)

Tangential stress (σ θ )

15

30

45

Distance from cave boundary, x(m)

60

σv:σh1:σh2=1:2:3 Undercut tunnel roof

7

In situ vertical stress (σ v)

0

60

6 5 4 3 hr= 50 m hr= 25 m hr= 15 m

2 1 0

15

30

45

Distance from cave boundary, x(m)

60

Figure 5.19: Predicted normalised tangential stresses in undercut drift roofs (Trueman et al 2002)

In general the magnitudes of the induced stresses increase with increasing hydraulic radius up to continuous caving. A doubling of the hydraulic radius leads to an approximately 30% increase in induced stress close to the cave front. Induced stresses are generally higher the closer the drift section is to the cave front and fall rapidly over the first 15 m away from the front. Drift sections therefore experience a gradual increase in induced stress as the cave front approaches. The magnitudes of the maximum induced stresses are, of course, higher on the undercut level than on the extraction level.

222

Chapter 5: Cave Initiation by Undercutting

5.5.6

Summary of Parametric Study Results

The results of the parametric study carried out by Trueman et al (2002) are generally in accord with the qualitative experiential guidelines summarised in Table 5.3. The study has permitted a number of these experiential guidelines to be quantified and bounded in the following ways. For extraction level drifts: •

as expected, the magnitude of induced boundary stresses in drifts is sensitive to the in situ stresses and their orientations;

•

if the hydraulic radius to achieve continuous caving doubles, the maximum induced stresses in the extraction level drifts increase by approximately 20% in most in situ stress environments. Exceptions to this occur in drift roofs where the major principal stress is approximately horizontal and perpendicular to the direction of drift advance;

•

when the vertical separation between the undercut and extraction levels is in the range of 10 m to 20 m, an increase or decrease in the separation distance of 5 m leads to an approximate 10% difference in induced stresses at drift boundaries; ie an increase in separation of 5 m leads to a 10% decrease and a 5 m decrease leads to a 10% increase in boundary stresses;

•

when continuous caving is achieved for a hydraulic radius of 25 m in a hydrostatic in situ stress field, the maximum induced stress reduces by 15% and the vertical induced stress reduces by 30% at the undercut level. In the extraction level drifts, the maximum induced stress reduces by 2% and the vertical induced stress reduces by 14%;

•

significant induced stress changes occur for many in situ stress states; ie stresses increase as sections of drifts are approached by the advancing cave and decrease as the cave passes overhead. Exceptions to this general rule occur in drift roofs in which the maximum in situ principal stress is approximately horizontal and perpendicular to the direction of cave advance;

•

in the main for an advance undercut, induced stress levels are significantly lower with the cave front 15 m in advance of the section of drift being investigated. This gives some credence to “the 45 degree rule”. Exceptions occur again in drift roofs where the maximum in situ principal stress is approximately horizontal and perpendicular to the direction of drift advance. Induced stress levels generally continue to fall, albeit at a reduced rate, more than 15 m from the cave front; and

223

Chapter 5: Cave Initiation by Undercutting

•

for an advance undercut, any excavations in advance of the cave front are subjected to induced stresses similar to those in a post-undercut. Only the excavations formed behind the cave front significantly benefit from an advance undercut.

For undercut level drifts: •

in general the magnitudes of the induced stresses increase with hydraulic radius up to continuous caving being achieved;

•

induced stresses are generally higher the closer the drift section is to the cave front, falling rapidly over the first 15 m away from the front. Exceptions are in drift roofs where the maximum principal in situ stress is horizontal and perpendicular to the direction of drift advance; and

•

the magnitudes of the maximum induced stresses are higher on the undercut level than on the extraction level.

5.5.7

Undercut Drift Support and Reinforcement

The provision of drift support and reinforcement forms an important part of undercut planning and design. Although they are not permanent excavations, the undercut drifts are vital components of an overall block or panel cave operation and must remain safe and accessible to men and machines throughout their design lives. As part of the International Caving Study, Stage I details were collected of a range of design and operational features of most currently operating caving mines. Summaries are given below of the support and reinforcement used in undercut drifts at four of these mines. Further details are given by Trueman et al (2002). Andina Panels II and III Panels II and III at CODELCO-Chile’s Andina mine were both extracted using a post- or conventional undercut strategy. Panel II was located entirely in a relatively weak rock mass known locally as secondary rock. The separation between the undercut and extraction levels was 15 m. The in situ stress regime was σv: σh1: σh2 = 9: 18: 13 MPa and the average Q' of the rock mass was 0.4, equivalent to an average RMRL of 35. Continuous caving was reported to have occurred at a hydraulic radius of 26 m. The average uniaxial compressive strength of the intact host rock was estimated to be 108 MPa. Panel III has a mixture of secondary (weak) and primary (moderately strong to strong) ore. Initial continuous caving was achieved in the weaker rock mass (having an average Q' of 0.4) at a hydraulic radius of 11.7 m. The in situ stress regime was σv: σh1: σh2 = 17: 22:13 MPa. The average uniaxial compressive strength of the intact host rock was estimated to be 108 MPa.

224

Chapter 5: Cave Initiation by Undercutting

Drift support and reinforcement was the same in both panels. In the undercut levels the drifts were generally reinforced only with spot bolting but wooden props were often erected in the immediate vicinity of the cave front. Generally drift conditions were good on both the undercut and extraction levels and, where placed, the support and reinforcement was adequate for the conditions in all sections of the drifts relative to the cave front. Esmeralda sector, El Teniente The Esmeralda sector of the El Teniente mine, Chile, is being extracted using a pre-undercut strategy (Rojas et al 2000b). The rock mass had a Q' of 5.3 up to continuous caving being achieved at a hydraulic radius of 27 m. The in situ stress regime was σv: σh1: σh2 = 26: 34: 34 MPa. The separation between the undercut and extraction levels was 18 m. The average uniaxial compressive strength of the intact host rock was estimated to be 100 MPa. The undercut level drifts are 4 m wide and 3.6 m high. These drifts are reinforced with 2.3 m long resin anchored bolts on a 1.0 m spacing and mesh. Generally the reinforcement is adequate, with damage being confined to local areas in which the rock mass is more fractured. Northparkes E26, Lift 1 The Northparkes E26, Lift 1 block cave used an advance undercut, with the extraction level and drawpoint drifts only being developed in front of the advancing undercut. The in situ stress regime was σv:σh1:σh2 = 12: 23: 15 MPa. The average Q' of the rock mass in the vicinity of the drifts was 8.7 (or an average RMRL of 53). The undercut was extended to a hydraulic radius of 44 m but continuous caving was not achieved and the cave was induced until it reached an overlying weaker rock mass. As shown in Figure 5.6 a double undercut was extracted. The lower undercut was developed between a fully developed upper undercut and the extraction level. The stress charts presented in Sections 5.5.4 and 5.5.5 do not take such a scenario into account. The average uniaxial compressive strength of the intact host rock was estimated to be 110 MPa. In the upper undercut, the drifts were 4.2 m wide by 4.5 m high. Installed reinforcement consisted of 2.1 m long split sets on a spacing of 1.25 m with 8 bolts per ring. In general, experience showed the reinforcement to be adequate. Palabora The Palabora block cave is being extracted using an advance undercut (Calder et al 2000). The average Q' of the main host rock was estimated to be 23 with the average RMRL being in the mid-70s. In situ stress measurements had indicated a hydrostatic state of stress of 38 MPa. The average uniaxial compressive strength of the intact host rock was estimated to be 140 MPa.

225

Chapter 5: Cave Initiation by Undercutting

The undercut level drifts are 4 m wide by 4 m high. Reinforcement consists of 2.4 m long resin grouted rebar on a spacing of 1.25 m. Additionally, weld mesh is installed with 1.0 m long split sets being used to secure the mesh in place. Fibre reinforced shotcrete nominally 50mm thick is used where ground conditions are considered to warrant it. Some spalling of the sidewalls is evident remote from the cave front even after bolts are placed but before mesh or shotcrete are installed. The shotcrete and mesh prevent further problems and the full support system appears to be working well close to the cave front at a hydraulic radius of 16 m.

5.6 5.6.1

DRILLING AND BLASTING FOR UNDERCUTTING AND DRAWBELL CONSTRUCTION Introduction

The requirement for proper design and accurate implementation of drill and blast strategies and practices should be stricter in block caving than in other underground mining methods. The consequences of poor drilling and blasting practices during block cave construction can be severe and can impact on the initial performance of a block cave (cave initiation) and the subsequent integrity of the extraction level drifts and drawpoints. Proper selection of blasting parameters such as hole diameter, explosive type and initiation sequence is therefore crucial. Just as important is the selection of drilling and blast hole charging equipment. The blasting parameters should be selected to suit the geomechanical properties of the rock mass to be blasted. The drilling equipment should be able to adequately and efficiently drill the required undercut, drift and drawbell geometries. The explosive loading equipment should be able to load the blast holes to design. This is particularly important when horizontal, inclined and uphole blast hole loading is required as is now common practice in block cave undercutting (see Section 5.4). Controlling blast performance and obtaining results to the level required during block cave construction was more difficult to achieve before the 1990s given the quality and reliability of some of the explosive products and accessories available at the time (Cameron and Grouhel 1990). However, the 1990s saw significant advances in commercial explosives technology including accessories. There is now available a wider range of gassed and variable density emulsions, low shock energy emulsions and ANFO/emulsion blends suitable for a wider range of hole diameters. In addition, very low velocity of detonation (VOD) products (eg diluted ANFO products and gassed emulsions) and highly accurate electronic and pyrotechnic detonators are now available. During the 1990s, significant advances were made in drilling equipment enabling easier alignment of underground drill rigs and more accurate drilling of holes. The technology and equipment for blast hole loading has also improved significantly as has understanding of rock breakage principles and explosive-rock interaction (eg Hustrulid 1999). With these developments, the opportunity now exists to “engineer” blasts to a level not feasible previously.

226

Chapter 5: Cave Initiation by Undercutting

During block cave construction the drilling and blasting objectives should be to ensure: • • • •

full breakage of the undercut rings without leaving remnant pillars; minimum damage to the perimeter walls during extraction level drift development; minimum damage to pillars during drawpoint construction; and achieving the required drawbell shape with minimum damage to the drawpoint brow.

This means that proper rock breakage design principles and rules should be applied during the design of undercut and drawbell rings and basic wall control principles should be applied during extraction level construction. 5.6.2

Factors Affecting Drilling and Blasting Performance

Through its wide research and field experience, the JKMRC has developed a number of experiential or empirical drilling and blasting guidelines that are supported by more fundamental investigations (Scott 1997). These have been implemented successfully at a number of current block caving operations (Guest et al 1995). The following sections discuss some of the design guidelines and rules that are applicable to block and panel cave drilling and blasting. Additional and practical principles that are also applicable to underground blasting are given by Hustrulid (1999). Rock parameters influencing blast performance The key intact rock and rock mass parameters that influence blast performance are: • rock density; • strength of intact rock blocks defined by the jointing; • jointing (intensity, orientation, persistence and condition); and • degree of attenuation and mode of failure. . In a given geological or geotechnical domain, any one or combination of the above parameters can have a significant influence on blast performance. The influence of these parameters is greater when blasting is carried out in confined environments such as undercuts and drawbells. In designing a blast for a given geotechnical environment, the engineer should assess which of the above factors are likely to control blast results and then take appropriate design measures such as those discussed below. Rock density Rocks typically blasted in mining have densities in the range 2.1 to 4.8. In blasting, rock density has a pronounced effect on the initial rock displacement or movement and the subsequent in-flight particle velocities or throw. For a given amount of energy, it is generally more difficult to break efficiently and displace higher density rocks. In terms of blast design, 227

Chapter 5: Cave Initiation by Undercutting

and where two or more rings are blasted at any one time, rock density should influence the choice of the inter-ring timing. Strength of intact rock blocks The strength of the in situ rock blocks isolated by the jointing is usually described in terms of either static or dynamic uniaxial compressive strength (UCS) values. In terms of blast design, the intact rock blocks should influence the density of holes drilled (S/B) and the characteristics of the explosive required. The explosive characteristics include, for example, detonation pressure, VOD, weight strength and brisance. Brisance is a property which describes the explosive’s ability to shatter a rock. Jointing As discussed in Chapters 2 and 4, the intensity, persistence and orientation of jointing define the sizes and distribution of the rock blocks that are either fully or partially formed in situ. The size and distribution of the in situ blocks determine the degree of breakage that might be required in order to achieve the desired fragmentation. Joint condition describes whether the joints are dilated, filled, closed or healed and in essence define joint strength. Joint strength determines how easily in situ rock blocks can be liberated during blasting. Healed joints can be as hard or harder to break as intact rock blocks. In terms of blast design, jointing and the corresponding size of the rock blocks formed should influence the characteristics of the explosives required (eg gas volume or VOD), the blast pattern (burden, B, and spacing , S,) and the design S/B ratio. The S/B ratio defines the density of holes which should provide the optimal distribution of explosive energy required to adequately fragment the in situ blocks. Where sub-optimal ratios are used, fragmentation can be structurally controlled with the resulting fragmentation distributions being equivalent to those of blocks in situ. In the case of high or double undercuts, achieving good fragmentation is important to ensure high loader productivity. Degree of attenuation and mode of failure An important rock parameter often ignored during blast design is the rock’s ability to transmit and absorb propagating blast energy and the rate at which that energy is absorbed. In seismology, the term attenuation is often used to describe this property. Different rocks can be assigned different attenuation constants. Highly attenuating rocks absorb propagating blast energy readily. There is a relationship between the rock matrix (grain size) and attenuation. For example, fine grained materials tend to be more highly attenuating; that is, the rate at which they absorb blast energy is greater than for rocks with a larger grain size (Chitombo 1991).

228

Chapter 5: Cave Initiation by Undercutting

Highly attenuating rocks also tend to fail “plastically” during blasting and low attenuating rocks generally fail in a more brittle manner. Rocks showing brittle behaviour generally transmit blast energy over relatively longer distances. The corresponding energy is likely to be evenly distributed resulting in even breakage. On the other hand, rocks that behave plastically, often referred to in mining as “soft” or “spongy”, rapidly absorb applied blast energy in the near vicinity of detonating blast holes. The resulting distribution of blast energy over the rest of the blast volume is poor and the breakage uneven. This is particularly the case where large blast hole spacings are used. Rocks showing brittle and plastic behaviours often require the application of different blast parameters to achieve the same degree of breakage. Soft, plastic rocks are generally more difficult to fragment adequately. In terms of blast design, the rock parameters which influence attenuation should also influence the choice of explosives (VOD or rate of energy release) and blast hole timing. The characteristics of an explosive are more critical in soft or weak rocks than in stronger rocks. 5.6.3

Experienced Based Design “Rules of Thumb” for Rock Breakage Control

The following sub-sections discuss design guidelines or “rules of thumb” that should be considered during drill and blast design to ensure optimal rock breakage. The rules presented are applicable to undercut blast design. Optimum S/B ratio Size of burden The empirical formulae reported in the literature for selecting a suitable burden (the distance between rings) for different rock masses, are mainly for unconfined, free-face blasting and not directly applicable to confined blasting. However, from experience and depending upon hole size, it is suggested that the burdens used in confined situations such as during undercutting should, as a general rule, be kept to a minimum. For small hole diameters (51 mm, 64 mm and 76 mm), the burdens should be in the range 1.2 m to 2.0 m, particularly when blasting soft/plastic type rocks. For larger hole diameters (eg 89 mm and 102 mm) the burdens should still be kept small with a suggested maximum of 2.5 m (based on experience). Burdens of 3.0 m have been used with the larger hole diameters but toe breakage problems are often experienced. Maximum toe spacing The choice of maximum blast hole toe spacing (S) should be governed by the size of ring burden (B) being blasted. For a given rock mass, there is an optimum S/B ratio that provides the optimum hole density, explosive energy distribution and therefore volume breakage. JKMRC experience is that the optimum S/B ratios applicable to a wide range of rock types and

229

Chapter 5: Cave Initiation by Undercutting

conditions are in the range of 1.1 to 1.4. This is consistent with figures given by Persson et al (1994). For soft, massive and/or sparsely jointed rock masses, the recommended S/B should be small and approaching 1.1. Under no circumstance should it be less than 1.0. S/B ratios of less than 1.0 tend to produce the so-called pre-split effect in which the breakage of the burden is poor. Ratios equal to 1.0 are not necessarily optimal in terms of the resulting breakage. Ratios of up to 1.3 have been found to be adequate for blasting some of the soft De Beers kimberlite rocks (Guest et al 1995) and the soft ultramafics rocks in Western Australia. For heavily jointed rock masses the S/B can approach 1.4. The influence of structure becomes prominent where ratios are equal to or greater than 1.4. When determining toe spacings, it is important that a consistent method of calculating toe spacing is used. Figure 5.20 shows examples of conventional methods of calculating toe spacing.

Figure 5.20: Conventional method for calculating toe spacing

230

Chapter 5: Cave Initiation by Undercutting

Choice of explosive type Based solely on the manner in which commercial explosives deliver energy into rock, the general explosive-rock interaction criteria shown in Table 5.5 can be used as a guide. Table 5.5: General explosive-rock interaction criteria Type of Rock

Explosive Characteristic Required

Hard (and brittle) Rock

Emulsions or High VOD Watergels

Soft (and plastic) Rock Very Soft Rock

ANFO or Heavy ANFO Diluted ANFO

The terms hard and soft rock relate not only to the intact rock strengths and mode of failure but also to the spacing and condition of the jointing. High VOD emulsion and water gel explosives have relatively high weight/bulk strengths and, more importantly, deliver their energy early and relatively quickly. They are therefore suitable for use with hard and brittle rocks. ANFO and heavy ANFOs tend to deliver their energy over longer periods. This makes them more suited to softer and more plastic rocks. However, before any pure emulsion or emulsion / ANFO blend type explosives products are considered, it is imperative that suppliers are asked to provide the following specifications for their products: • • •

•

the critical diameters of the products; sensitivity of the products in the hole sizes to be used. This is particularly important for small diameter holes; for emulsions and emulsion blends sensitised using micro-balloons, the likelihood of the products being shock desensitised should be quantified. This is particularly important given the size of hole diameters and the relatively small blast hole spacings being currently applied; and the VOD versus hole diameter relationships of the products (including ANFO).

Any one of these factors can affect efficient detonation of the products, inevitably influencing blast performance and results. Average powder factors (kg/t) and corresponding energy distributions During undercutting, the optimum powder factor (kg/t) is that which ensures full and proper breakage of a ring, in particular the “toe” and “shoulder" sections. However, the breakage of the full blast volume is dependent upon the distribution of the explosive energy and not necessarily on the average powder factor. The concept of explosive distribution is discussed later.

231

Chapter 5: Cave Initiation by Undercutting

In practice, both the average powder factor and the optimum explosive distributions are obtained through trial and error methods. To speed up this optimization process, the JKMRC has developed software (JKSimBlast) to calculate the average powder factor and the corresponding energy distribution. Energy distribution is dependent upon rock type, hole diameter, the S/B ratio, explosive type and timing. Experience in blasting soft and plastic kimberlite rock at the Kimberley and Koffiefontein Mines in South Africa, showed that a powder factor of 0.35kg/t at the toe ensured full breakage of relatively confined undercut rings. During the mining of the Northparkes E26 Lift 1 block cave, powder factors in the range 0.25kg/t – 0.3kg/t provided the required undercut ring breakage. In most hard rock and confined blasting situations, powder factors in the range 0.25kg/t – 0.35kg/t are generally recommended. The charging geometry is also important in achieving the required explosive distribution. A recommended charging configuration and rule is given in Figure 5.21. The recommended charging also ensures that the concentration at the collar is reduced thereby minimizing the potential for brow damage.

Combination of patterns used to distribute the charge

Figure 5.21: Recommended blast hole charging patterns

232

Chapter 5: Cave Initiation by Undercutting

Blasthole timing The use of high powder factors does not necessarily ensure the efficient breakage of a volume of rock. Efficient breakage of a blast volume is better promoted by the manner in which blast energy is distributed and imparted into the rock. This is controlled by blast hole timing. Single hole or hole-by-hole firing with an inter-hole timing in the range 17 to 30 ms is generally recommended for improving breakage in most hard rock blasting. This ensures that the energies from adjacent holes contribute to the breakage of the volume of rock straddled by the adjacent holes before displacement. With electronic or any super accurate system where overlapping of delays is unlikely, the inter-hole timing can be in the range 9 to 25 ms. In general, faster times are more suited to hard and brittle rocks and slower times to more “plastic” and high density rocks. Firing ring holes on the same delay (cluster blasting) has been shown through experience to produce poor breakage particularly in soft rocks. In such cases the fragmentation is generally uneven, consisting mainly of fines and coarse fragments in a bi-modal distribution. Cluster blasting in not a recommended practice during undercutting or drawbell blasting. This method of blast hole firing may also cause damage to the adjacent rock. The influence of timing on fragmentation is discussed in a paper on the evolution of compound rings by Guest et al (1995). Based on the JKMRC’s experience in blasting TKB kimberlite, the inter-hole delay should be less than 40 ms. Analysis of high speed films taken at Koffiefontein and Finsch mines in the early 1990s showed that 3.2 m burdens using 89 mm holes charged with an emulsion product, moved after 40 ms to 50 ms of detonation. This means that if delay times of ±40 ms are used, then the adjacent holes will work independently resulting in poor fragmentation and an increased likelihood of hole cut-off. Historically, it has been difficult to control fragmentation because of the scatter or dispersion associated with delay elements. However, there has been significant improvement in pyrotechnics and the delay scatter in current systems is minimal thus making it possible to better control blast fragmentation. In selecting delay elements it is important that suppliers are asked to provide relevant performance data. Hole diameter The selection of hole diameter is generally influenced by the available equipment and, more importantly, by the lengths of the holes to be drilled. For the hole length generally drilled in most block cave undercuts (ie less than 30 m), 64 mm, 76 mm and 89 mm are suitable diameters with 102 mm diameter holes considered the recommended maximum. Larger diameter holes also mean that the potential for increased powder factors and kilograms of explosive per delay, which increases the potential for blast damage.

233

Chapter 5: Cave Initiation by Undercutting

5.6.4

Undercut Drilling and Blasting

As has been discussed earlier in this Chapter, the successful development and extraction of an undercut is essential for the initiation of the caving process. The best drilling and blasting practices must be implemented to avoid poor breakage resulting in the formation of remnant pillars (Figure 5.22) which not only inhibit the initiation of caving but also act as loading points or “stress channels” thereby increasing the likelihood of damage to extraction level drifts, drawpoints and support. The planned geometry of an undercut should govern the drilling and blasting parameters and practices to be used. The following criteria should be considered during design: • • • • • •

ease of drilling to ensure proper distribution of holes required to completely break the rock volume thereby avoiding the formation of pillars; ease of loading or clearance of blasted material to avoid blasting of subsequent rings under severe choked conditions; potential for hole losses or closures during undercutting due to either blast damage, stress abutment effects or relaxation of the rock mass; ease of managing the undercutting sequencing and achieving the required undercut front shape; problems likely to be encountered during undercutting and ease of recovery from such problems; and potential for large failed blocks reporting early into the blasted undercut immediately following undercutting.

The types of undercut geometries used have been discussed in Section 5.4. Regardless of the type of undercut geometry used, the following critical drilling and blasting factors should be considered: • • • •

height of undercut; hole inclination; amount of explosives or explosive energy distribution; and amount of blasted material to be loaded.

Height of undercut The influence of the height of the undercut on undercut and caving performance was discussed in Section 5.3.5 and examples of undercut heights used, in particular at El Teniente, were given in Table 5.2. Purely from a breakage view point, higher undercuts are easier to break because of the increased free-face area. The potential for not achieving full breakage is greater in narrow undercuts because of confinement. Any small amount of hole deviation can also exacerbate the confinement problem.

234

Chapter 5: Cave Initiation by Undercutting

Figure 5.22: Examples of remnant pillars A number of block caving operations have encountered significant drilling and blasting problems during the mining of narrow undercuts. In Esmeralda, El Teniente, changes to the drilling geometries were made to improve breakage as shown in Figures 5.9 and 5.10. Bell Mine increased the undercut height from approximately 3.2 m to 6.0 m also as a way of improving breakage (Figure 5.23).

235

Chapter 5: Cave Initiation by Undercutting

Figure 5.23: Undercut geometry at Bell Mine, Canada Hole inclination and ring geometry Narrow flat undercut For a narrow flat undercut, parallel or near parallel holes are considered ideal. However, the ability to do this will depend upon the capabilities of the drilling equipment available. Where the drilling equipment is unable to drill parallel holes, then a fan from a single drilling point can be utilized. These alternative drilling geometries are shown in Figure 5.24.

Figure 5.24: Typical drilling geometries for a narrow flat undercut

236

Chapter 5: Cave Initiation by Undercutting

Parallel holes provide the best explosive energy distribution. Fan drilling from a single point will require slightly more holes to provide the same explosive distribution. To achieve the required throw with either drilling configuration the holes should be angled slightly forward. From experience the recommended angle range is 20° to 30°. Uphole or SLC undercut rings The recommended practice is to slightly dump the uphole rings forward by 10° or 15°. Toe breakage has been demonstrated in practice to be better with the slightly forward dumped rings. Vertical or 90° rings often result in “crown formation”. The impact of hole inclination is well illustrated in the Block Cave Manual (Laubscher 2000) by the work of Bell at Shabanie Mines (Figure 5.25). Planned limit of break

Area unbroken just shattered Probable break outline

Broken ground

Direction of retreat

Development

Undercut with 90 degree rings Planned limit of break

Area unbroken just shattered Probable break outline

Broken ground

Direction of retreat

Development

Undercut with 70 degree rings Figure 5.25: Impact of ring inclination (Laubscher 2000)

237

Chapter 5: Cave Initiation by Undercutting

Explosive energy distribution The powder factor is an index often used to determine the amount of explosive required to adequately fragment a mass or volume of rock. It is simply the number of kilograms of explosive required per tonne or cubic metre of rock to be broken (ie kg/t or kg/m3). In practice, engineers experimentally determine a range of powder factors or the average powder factor required to achieve optimum breakage and fragmentation. Powder factors to suit a given rock mass can also be approximated using blastability indices (eg Lilly 1986). Unfortunately, an average powder factor assumes that the explosive energy is uniformly distributed within the blast volume. While this may be a good approximation in a bench or parallel hole geometry, the concept is not strictly correct in the case of a ring or fan blast hole geometry. The JKMRC has introduced the concept of the three dimensional explosive energy distribution as a better way of accounting for the explosive energy within the rock volume to be blasted (Kleine 1988). The methodology considers that each point in the rock will have an energy value arising from each detonating blast hole, determined by the distance between that point and all detonating holes and the amount of explosive in each hole. With reference to Figure 5.26, the powder factor calculation was extended by considering a small infinitesimal segment of charge and writing the equation for the resulting explosive concentration at a point P for a sphere centered at the charge segment. The general form of the equation is:

P=

⎛D⎞ ⎟ L 2 1000.ρ e .π⎜ ⎝2⎠

∫

L1 ρ

r

(

4 π h 2 + l2 3

2

)

2 3

(5.1)

dl

Equation (5.1) can be integrated and rewritten as:

P = 187.5

ρe 2 1 D ρr h2

⎛ L 2 L1 ⎞ ⎟ ⎜⎜ − r1 ⎟⎠ ⎝ r2

(5.2)

Special conditions apply to the above relationships at the charge axis (ie h = 0) and at very large distances (ie h Æ∞). The explosive concentration at any point in three dimensions is determined by solving the appropriate integrated form of the equation for each explosive charge and summing the values.

238

Chapter 5: Cave Initiation by Undercutting

Figure 5.26: Calculation of the 3D explosive energy concentration at point P The explosive energy concentration calculated by this method will generally be slightly higher than the conventional powder factor. However, the values will be similar at the blast’s designed burden. The highest concentration of energy is in and around each blast hole. The JKMRC blasting software incorporates this model. The energy distribution can be calculated in any plane in three dimensions. This approach has been applied successfully to undercutting at the De Beers caving operations and at the Northparkes E26 Lift 1 block cave. In both instances, the undercuts were successfully mined without leaving pillars. Figure 5.27 shows an example of the analysis used to calculate the distribution of explosives in some of the undercut rings at Northparkes. From the analysis it is possible to assess the effect of increased burden or number of holes. The idea is to achieve the required powder factor in the toe regions in the case of fans.

239

Chapter 5: Cave Initiation by Undercutting

Reference undercut ring - NORTHPARKES Lift 1

Toe Spacing = 3m, Explosive = Emulsion 1 g/cc Rock SG = 2.8

3D Explosive Energy Distribution

1.6 m Burden

2.0 m Burden

2.5 m Burden

Figure 5.27: Use of the 3-D explosive energy distribution to optimise ring design

240

Chapter 5: Cave Initiation by Undercutting

Amount of blasted materials to be loaded Conventional and recommended practice during undercutting is to remove some material to avoid blasting under “choked” conditions. As a general rule, 20% to 40% of the blasted volume is removed which is equivalent to removing the “swell”. This ensures that the subsequent rings are not blasted under full confinement. Laubscher (2000), however, suggests that in the case of narrow flat undercuts, sufficient material should be left to support the backs thereby reducing abutment stresses. This is analogous to the effect of backfilling stopes in narrow gold reef mining, a practice that has been shown to decrease abutment stresses. Experience has shown that when excessive material is removed, abutment loading can result.

5.6.5

Drawbell Blasting

Figures 5.28 and 5.29 provide examples of some of the drawbell shapes currently used. In designing the most optimal drawbell shape and size, the following drilling and blasting issues should be considered: •

ease of drilling the patterns considered optimal for achieving the bell size and shape required to induce flow;

•

the ability to shape the major apex pillar so as to minimize the width of the apex ;

•

the practicality of drilling wider fan holes required to shape the drawbell to the preferred dimensions from the relatively short base of the drawbells; and

•

ease of charging the rings to achieve the required powder factors and explosive energy distributions for efficient drawbell blasting without excessive charging of ring collars given the concentration of holes.

It is also recommended that the number of blasts or stages used to open a drawbell be minimised. An added benefit of the reduced stage drawbell opening sequence is increased safety for operators when charging. The time spent within the blasted drawbell area is significantly reduced. The impact of vibrations from the different stages is also minimised. The recommended stages are: • raising; • slotting; • stripping of fans into the slot on one side; and • stripping of fans into the slot on the second side.

241

Chapter 5: Cave Initiation by Undercutting

A common and recommended drawbell raising practice is “blind raising” using large diameter (660 mm and 1200 mm) holes drilled using equipment such as the RD 2000 from Master Drilling. The recommended hole diameters for drawbell blasting are 64 mm and 76 mm with 89 mm being the recommended maximum size. A recent innovation in drawbell blasting was made at the Premier Mine’s BA5 block cave where the drawbells were blasted using electronic delay detonators. This enabled hole by hole firing which effectively ensures that each hole is primed and detonated and that the resulting amount of explosive detonated per delay is greatly reduced. This is important for blast damage considerations.

10m

5.2m

Raise Stage 1- Stripping

3.6m

Stage 2- Slot

Strip A

Slot

Strip B Stage 3 and 4- Strip A & B

5.2m

14m Figure 5.28: Skull shape drawbell for Northparkes E26 Lift1 block cave and the blasting sequence used

242

Chapter 5: Cave Initiation by Undercutting

Excavation 1° - 2° Phase

Cross Section Figure 5.29: Drawbell shapes and sizes used for the Esmeralda section, El Teniente Mine, Chile

5.6.6

Drilling Equipment Selection

Proper selection of drilling equipment is crucial for producing the required undercut, drawpoint and drawbell shapes. The selection criteria should include: • • •

size of unit; boom coverage; and ease of drilling horizontal and angled holes.

243

Chapter 5: Cave Initiation by Undercutting

Almost all major equipment manufacturers now have or can configure rigs to meet most requirements for undercut and drawbell drilling. Examples of rigs that can be considered for block cave applications are described below.

Mercury 1LC10 (D4 – E50) This is a high power compact top-hammer long hole drill, with a four wheel drive articulated hydraulic jumbo, for fully mechanised production from medium sized drifts. This Tamrock rig can be easily operated in 4.2 m wide drives and has a good boom coverage because of its flexible boom head. This means that horizontal holes can be drilled more easily and closer to the floor. Standard 1.8 m rods can be used. The rig is capable of drilling 76 – 89 mm holes when equipped with the Hydrastar 300 SR.

Solomatic 620 The Tamrock Solomatic 620 is a flexible one boom electrohydraulic long-hole rig for production drilling underground. The boom has a wide parallel drilling coverage, long boom extension, 360° rotation and wide tilt angle ranges forwards and backwards, thus offering wide drilling variety. The rig is capable of drilling 64 – 89 mm holes up to 32 m long. This is more suited to wider tunnel sizes, nominally 4.5 m, if conventional 1.8 m rods are to be used. If this unit is to be used without modification, then undercut drives have to be mined to a width of 4.5 m. Other suitable rigs include the Atlas Copco Simba 1354 Series. It is important that the rigs are adequately equipped with appropriate angling devices to permit the proper drilling of angled holes.

Accessories Rig and hole alignment systems now exist that can be mounted on drilling rigs. One such device is the Transtronic angle indicator type 6H. The system can be installed in other types of rigs. It comprises two gravity type sensors that measure boom inclination and feed rotation. The device can measure side angle measurements to 360° with an accuracy of ± 0.2°. The inclination angle measurement range is ± 60° with an accuracy of ± 0.8°. The application of tubes is also recommended for accurate hole drilling. Tubes are now available to suit 76 mm and 89 mm diameter holes.

244

CHAPTER 6 EXTRACTION LEVEL DESIGN

6.1

PURPOSE

T

he operational efficiency and cost effectiveness of block and panel caving mines depend, among other things, on the design and performance of the extraction or production level excavations. As earlier discussions have indicated, extraction level design and performance are influenced by the degree of fragmentation achieved and by the undercutting strategy adopted. The three-dimensional geometries of the excavations between the extraction and undercut levels can be very complex. They must be designed to ensure that they remain stable and conducive to efficient production operations throughout their design lives which can be several years, or even decades in some cases. As in the example shown in Figure 1.9, the early block caving mines used gravity loading systems via grizzlies and a combination of finger and transfer raises to the haulage level. This system is best suited to ore that fragments finely but is labour intensive requiring significant development. In other cases, including some of the South African diamond mines, a slusher system was used to transfer the ore from the drawpoints to the haulage. A good account of gravity and slusher draw systems is given by Pillar (1981) and summarised by Brady and Brown (1993). An example of a slusher draw system is shown in Figure 6.1. Although grizzly and slusher systems still find some use, they have been almost completely replaced by mechanised methods of drawing and moving the ore on the extraction level using Load-HaulDump (LHD) vehicles. Accordingly, extraction level layouts for only mechanised methods of loading will be considered here. The purpose of this chapter is to discuss the factors influencing extraction level design and performance, the advantages and disadvantages of a number of types of extraction level layout, the design of drawbells and drawpoints and the stability, support and reinforcement of extraction level installations. Consideration will be restricted to horizontal extraction level layouts. Discussions of a range of inclined drawpoint layouts are given by Carew (1992), Jakubec (1992), Laubscher and Esterhuizen (1994) and Laubscher (2000).

245

Chapter 6: Extraction Level Design

Figure 6.1: Slusher draw system (after Pillar 1981)

6.2

FACTORS INFLUENCING EXTRACTION LEVEL DESIGN AND PERFORMANCE

In order to introduce what can be a complex topic, the major factors influencing extraction level design and performance will be listed and discussed briefly before more detailed consideration is given to some of them later in this chapter. Discussions of many of these factors are given by Esterhuizen and Laubscher (1992) and Laubscher (1994, 2000).

246

Chapter 6: Extraction Level Design

Fragmentation The fragmentation of the ore produced in the draw column influences the choice of draw system used. Gravity systems require finely fragmenting ores while LHD systems are the natural choice for the more coarsely fragmenting ores commonly encountered in modern block and panel caving mines. The degree of fragmentation determines the size of the draw zone and hence the drawpoint spacing. It also influences the height of the drawpoint, the need for access for secondary breaking, the shape of the major apex, the LHD size and crushing requirements. Undercut Strategy and Design As has been discussed in Chapter 5, the undercutting strategy adopted (post-, pre- or advance undercut) influences the stresses induced in extraction level excavations, the need for support and reinforcement, the rate at which drawpoints can be brought into production and their longterm performance. The design and performance of the extraction level excavations are also influenced by the detailed design of the undercut as discussed in Section 5.4. For example, the undercut shape can influence the tendency for ore to stack and induce excessive loads on the pillars between the drawpoints (see Figure 5.11). Geotechnical conditions Because the percentage of excavation on and above extraction levels is so high, and the performance of the excavations is critical to the continuity and efficiency of production, it follows that the prevailing geotechnical conditions have major influences on extraction level design and performance. Depending on the undercut strategy and design adopted, the stresses induced in the excavations can be expected to be high and to change throughout the history of development, undercutting and production. The geotechnical characteristics of the rock masses discussed in Chapter 2 (eg major and minor discontinuities, intact and rock mass strengths) and their relation to the in situ and induced stresses will be major factors to be taken into account in the design. They can influence the sizes, shapes and the need for support and reinforcement and repair of the excavations. These issues will be discussed in more detail in Section 6.5. Operational factors The extraction level layout must be conducive to the efficient removal of the broken ore. The ease and speed of development must also be taken into account in design. A complex layout may require twice as much time to develop as a simple layout (Esterhuizen and Laubscher 1992). The sizes, shapes and geometrical relationships of the extraction level excavations must allow ease of access of LHDs to drawpoints, loading, reversing, turning into and out of drifts, travel to unloading points and unloading. Access of men and machines to deal with hangups in the drawbells or stacking on the major apex must also be considered.

247

Chapter 6: Extraction Level Design

Major operational hazards The major operational hazards to be discussed in Chapter 10 – major rock falls, rock bursts, mud rushes, air blasts and water and slurry inflows – can all have significant impacts on the extraction level. The prevention of their occurrence is the preferred way of alleviating the effects of these hazards. In some cases, the provision of appropriately designed support and reinforcement can alleviate the impact of rock falls, rock bursts and air blasts on extraction level installations. These issues will not be considered further here but will be discussed in Chapter 10. Drawpoint brows As in some other mining methods the stability, reinforcement, wear and repair of drawpoint brows can have major influences on the efficiency of production operations in block and panel caving mines. The issues associated with drawpoint brows could be considered to form part of the other design and geotechnical considerations outlined above but they are considered to be so important that they have been listed here separately. Major influences on brow stability and performance are brow orientation, the presence and orientation of discontinuities (because the brow shape provides release surfaces not present in other excavations), the impact of stress abutments during undercutting, the nature of the reinforcement used and the timing of its installation, and brow wear or deterioration during production.

6.3 6.3.1

EXTRACTION LEVEL LAYOUTS Scope

Laubscher (2000) has identified 10 different horizontal LHD layouts as having been used in block caving mines. Given that some are local variants of the major types, only five major types of layout will be considered here. In this context, the layout will be taken to refer to the arrangement of production drifts, cross-cuts or drawpoint drifts and drawpoints on the extraction level. The layouts to be considered here are the continuous trough, herringbone, offset herringbone, Henderson or Z design, and parallelogram or El Teniente layouts, and some of their variations. The issues of ore passes, separate haulage levels and crusher locations are discussed generically in Section 6.3.7. The discussion is based largely on those of Esterhuizen and Laubscher (1992) and Laubscher (2000). 6.3.2

Continuous Trough or Trench Layout

Weiss (1981) describes the continuous trough or trench layout used under high stress conditions at the Austro-American Magnetite Company’s mine in Austria. This layout has also been used at the Shabanie Mine, Zimbabwe (Laubscher 2000), the San Manuel Mine, USA (Stevens et al 1987) and considered at El Teniente (Jofre et al 2000). Figure 6.2 shows the layout used at the

248

Chapter 6: Extraction Level Design

Austro-American Magnetite Company’s mine. The longitudinal undercut trough drift was developed and blasted before the associated production drifts and drawpoints on the same level. This system involves no minor apices.

1 = ore, 2 = caved in area, 3 = surrounding rock, 4 = head slot, 5 = slots, 6 = drawing drift, 7 = fan drills, 8 = strike drift, 9 = main hauling drift, 10 = drill holes for dewatering

Figure 6.2: Continuous longitudinal trough layout used at the Austro-American Magnetite Mine, Austria (Weiss 1981) Weiss (1981) suggests that this layout aided the stability of the major apex under the prevailing high stress conditions, although as Laubscher (2000) points out, there is no lateral restraint to the major apex as is normally provided by the minor apices. The lower percentage of excavation on the extraction level than in other layouts should help avoid some stability problems. Selection of the optimum orientation of the trough with respect to the major horizontal principal stress direction is especially important with this layout. If the undercut trough is elevated above the extraction level then a small minor apex is created. Twodimensional numerical stress analyses carried out by Jude (1990) for the San Manuel trial showed that an elevated trough resulted in the elimination of a destressed zone which developed in the non-elevated case and so helped maintain a compression arch over the production drift. A disadvantage of the continuous trough layout is that the drawpoints associated with a given production drift use the same draw cone so that problems with the cone can affect production from several drawpoints. 249

Chapter 6: Extraction Level Design

6.3.3

Herringbone Layout

A herringbone layout with directly opposite drawpoints as illustrated in plan view in Figure 6.3, has been used at King and Shabanie mines, Zimbabwe (Laubscher 2000, Wilson 2000). In this case, the minor apices are in line with each other and at right angles to the major apex. Compared to other layouts of similar dimensions (see Figures 6.4, 6.6 and 6.7, for example), this produces unfavourable stability conditions because of the two acute corners (which will be rounded in practice) opposite each other and the large effective spans created by the opposing cross-cuts (Esterhuizen and Laubscher 1992). Operationally, this layout does not have the advantage that a loaded LHD can back readily into the opposite cross-cut or drawpoint drift for ease of turning. This can have safety implications in mud rush prone mines, for example. For these various reasons, the original herringbone layout has been modified to improve its performance.

Figure 6.3: Typical herringbone layout analysed by Esterhuizen and Laubscher (1992) 6.3.4

Offset Herringbone Layout

In the offset herringbone layout illustrated in Figure 6.4, the drawpoints on opposite sides of a production drift are offset or staggered. This helps improve both the stability conditions and the operational efficiency over those applying in the symmetrical herringbone layout discussed above. This system was used initially at the Henderson Mine, USA, and at the Bell Mine, Canada. It has become the layout most commonly adopted in the newer block and panel caving operations including Northparkes (Duffield 2000), Palabora (Calder et al 2000), Premier BB1E and C-cut (Bartlett and Croll 2000) and Freeport Indonesia’s Deep Ore Zone (Barber et al 2000).

250

Chapter 6: Extraction Level Design

Figure 6.4: Typical offset herringbone layout analysed by Esterhuizen and Laubscher (1992) A plan of the full extraction level layout adopted at Palabora is shown in Figure 6.5. Calder et al (2000) report that this layout was chosen principally on the basis of LHD manoeuvrability and the potential that it provided to use electric LHDs with trailing cables. With the crushers located on one side of the production area, all of the drawpoints point towards the incoming LHD. In addition, with the return ventilation on the opposite side of the production area to the crushers, any dust from the loaded LHD bucket is coursed away from the LHD driver. It is instructive to note in this regard that the Northparkes E26 Lift 1 block cave used crushers on opposite sides of the production area but that the layout for Lift 2 has been simplified through the use of crushers on one side only (Duffield 2000).

Figure 6.5: Extraction level layout, Palabora Mine, South Africa (Calder et al 2000)

251

Chapter 6: Extraction Level Design

6.3.5

Henderson or Z Layout

In this design illustrated in Figure 6.6, the opposite drawpoint drifts are in line but inclined to the production drift, and the drawpoints and drawbells are at right angles to the production drift. The minor apices are again in line and at right angles to the major apex but, in this case, the acute corners are diagonally opposite each other. This layout was first used at the Henderson Mine, USA, and so is sometimes referred to as the Henderson layout. Interestingly, it is not being used in the newer parts of the Henderson Mine where the simpler layout shown in Figure 1.16 has been adopted. No other current cases of the use of this layout are known.

Figure 6.6: Typical Henderson layout analysed by Esterhuizen and Laubscher (1992) 6.3.6

El Teniente Layout

In the layout developed at the El Teniente Mine, Chile, illustrated in Figure 6.7 (with a more detailed recent example shown in Figure 1.11), the drawpoint drifts (called zanjas) are developed in straight lines oriented at 60o to the production drifts (or calles) and the major apices. The minor apices are short and inclined to the major apices. Figure 6.7a shows the original layout with square drawbells at right angles to the drawpoint drifts. Figure 1.16 shows a similar, but not identical, layout being used at the Henderson Mine, USA, in which the angle between the production and drawpoint drifts is 56o and the drawpoint is divided into two by a small pillar (Rech et al 2000). Figure 6.7b shows one of a number of revised layouts developed at El Teniente in response to a range of mining conditions (Jofre et al 2000). The drawbell shape shown in Figure 6.7b was developed to increase the interaction between drawpoints and to improve downhole blast design.

252

Chapter 6: Extraction Level Design

Figure 6.7: (a) Typical El Teniente layout analysed by Esterhuizen and Laubscher (1992), and (b) modified El Teniente layout with decahedral drawbells (Jofre et al 2000) Numerical modelling carried out by Esterhuizen and Laubscher (1992) showed the El Teniente layout to be “stronger” than the comparable herringbone, offset herringbome and Henderson layouts. The minor apex in the El Teniente layout is shorter and more stable than those in the other layouts. The orientation of the layout with respect to the major horizontal principal stress direction has a critical influence on the performance of the El Teniente layout in high stress conditions. The most favourable stress conditions are likely to be achieved when the major principal horizontal stress is oriented parallel to the production drifts. An advantage of the El Teniente and new Henderson layouts is that the LHD can back into the opposite drawpoint drift for turning or for straight-on loading when there is brow wear. The only real disadvantage of the layout is that it is not suitable for electric LHDs using trailing cables (Laubscher 2000). 6.3.7

Ore crushing and transportation

Although ore crushing and transportation are not central to the concerns of this book, they will be discussed briefly here for completeness. As illustrated in Figure 1.9, the early block caving mines generally used finger and transfer raises to transport the broken ore from the grizzly to the haulage level where it was loaded onto a tracked transport system. Underground crushers were not required for many of the weak, finely fragmenting ores for which this mining method was developed. Different approaches are required for medium and coarsely fragmenting ores and for the mechanised loading of ore. A modern example of the Henderson Mine, USA, is summarised in Section 1.3.4 and illustrated in Figures 1.14 and 1.15. Full details are given by Rech et al (2000). In this case, the ore is transported by LHDs to ore passes on the extraction level and then transferred through the ore pass system to the haulage level 194 m below. The ore is then loaded into side-dumping trucks and hauled to the crusher dump on this level. It then passes down to the crusher and, after crushing, is transported out of the mine and to the mill by a very long conveyor system.

253

Chapter 6: Extraction Level Design

A number of recently developed block and panel caving mines use more compact layouts. As noted in the discussion of the offset herringbone layout in Section 6.3.4, the Northparkes E26 Mine Lift 1 uses crushers on the extraction level on opposite sides of the production area. From the crushers, the ore is transported to surface via a conveyor system and a hoisting shaft as illustrated in the schematic vertical section shown in Figure 6.8. The deeper Lift 2 will use a single crusher on the extraction level and an inclined conveyor system to the existing hoisting shaft (Duffield 2000).

Figure 6.8: Schematic vertical section, Northparkes E26 Mine (Dawson 1995) Figure 6.5 shows a plan of the extraction level layout for the larger Palabora block cave. Here, four crushers are installed on one side of the production area to optimise LHD travel distance and provide flexibility if a crusher becomes unavailable (Calder et al 2000). The crushed ore is then transported via an inclined conveyor system to the hoisting shaft. The planned block caving operation at Bingham Canyon, USA, will use a different system again. In the two planned block cave areas, LHDs will muck ore from drawpoints to ore passes which will then feed the ore to a wide heavy-duty conveyor belt. The conveyor will transfer the ore to a central crusher. The crusher will discharge onto a series of long inclined conveyors which will deliver the crushed ore to the existing surface stockpile (Carter and Russell 2000). With the increasing sizes and speeds of LHDs, and the introduction of tele-operated and fully automated machines, the quality of extraction level roadways has become of increasing importance in caving mines. As well as helping improve productivity and reduce LHD maintenance costs, good roadways can also help manage problems of water ingress and contribute to the development of a closed support ring in high stress conditions as discussed in 254

Chapter 6: Extraction Level Design

Section 6.5.2. The quality and performance of roadways are influenced by their design, construction and maintenance, and by the treatment that they receive during production operations. There are differing views in the industry about the value of certain roadway types and designs, the levels of expenditure merited on roadways and the desirable forms and standards of roadway maintenance. Some of these issues are canvassed by Laubscher (2000). While this is not the place to explore these issues in detail, a summary of the advantages and disadvantages a range of roadway types prepared by N J W Bell is reproduced in Table 6.1.

6.4 6.4.1

DRAWPOINT AND DRAWBELL DESIGN Gravity Flow of Caved Ore

Drawpoint and drawbell design are related to, among other things, the degree of fragmentation of the ore and its flow characteristics. Although the topic has been studied almost since caving methods of mining were introduced (eg Lehman 1916), the gravity flow characteristics of caved ore are still not well understood. Based on these studies and accumulated mining experience, a number of principles and guidelines to drawpoint spacing and design can be offered although a formulaic approach is still not available. What has come to be regarded as the classical approach to describing the gravity flow of ore involving the concept of the flow ellipsoid, was developed and applied initially to sub-level caving by Kvapil (1965, 1992) and Janelid and Kvapil (1966). Although it is known to have some deficiencies (Just 1981, Rustan 2000), this approach will be adopted here for purposes of illustration. A useful summary of this approach has been given by Otuoyne (2000). On the basis of a range of laboratory and field tests, it was postulated that if the ore is contained in a bin or bunker and a bottom outlet is opened, the material that will have been discharged after a given period of time will have all originated from within an approximately ellipsoidal zone known as the ellipsoid of motion, draw or extraction (Figure 6.9). Material between the ellipsoid of motion and a corresponding limit ellipsoid or loosening ellipsoid will have loosened and displaced but will not have reached the discharge point. The material outside the limit ellipsoid will remain stationary. As draw proceeds, an originally horizontal line drawn through the broken material will deflect downwards in the shape of an inverted “cone”. The shape of this draw cone indicates how the largest displacements occur in a central flow channel.

255

Chapter 6: Extraction Level Design

Table 6.1: Advantages and disadvantages of some extraction level roadway types (Laubscher 2000) ROADWAY TYPE Conventional concrete

ADVANTAGES A good finished surface that can be well controlled to give a strong floor in intermit contact with the hard rock beneath. Can be reinforced with weld mesh panels and protected from LHD digging by inclusion of rails.

LHD interlocking roadway bricks

Compressed bricks – e.g. G pattern, cubic or other varieties. Reinforced concrete with rails/RSJ’s

Roller compacted concrete

Steel plate in the loading area

Run of mine gravel

Readily laid on a compacted floor, which does not have to be dug to solid. No delays to travel. Readily repairable in case of damage e.g. floor lift. Concrete is guaranteed to be of good strength to stand the wear and tear of the LHD travel. Same as the interlocking brick, except easier and cheaper to make. As for concrete floors, however the rails have to be put in place before hand and this increases the costs. However, it does improve the wear characteristics and maintenance. These are particularly recommended for loading areas where the wear and tear of the bucket into the floor, if not done, can be horrendous leading to large holes being dug in the floor. If the logistics can be overcome and controls put in place this could well be a very useful material for initial floors and for repairs of existing concrete floors or other floors. These have been successfully used in development and trials are to be conducted at King in the next set of draw points to see if this idea has merit. Easy to lay.

256

DISADVANTAGES Requires a minimum of seven days curing time with no traffic on it, so the end is not available for other work during this time and the floor should perforce be dug out to solid or the material left on the floor properly compacted before the final concrete is poured. Repairs are very difficult as the concrete has to be dug out and a further seven days waited for the replacement concrete to cure. Manpower to install the bricks. The costs of the bricks and their moulds etc, which are very high. Can’t be used in drawpoint loading areas without concrete or steel plate.

Same as the interlocking brick, but as laborious if not more so to place and tend to tear out on corners. If footwall heave occurs the whole floor lifts and access into the area can be prevented or huge damage to tyres results.

Underground this has proved difficult to install as the controls have to be exceptional to make the concrete to the required strength.

High rolling resistance to vehicles. Maintenance costs ongoing and higher. Material must be suitable.

Chapter 6: Extraction Level Design

Figure 6.9: The flow ellipsoid concept of the gravity flow of broken ore (Kvapil 1992) A number of more recent field and laboratory studies have shown that the “ellipsoid” is not always a true ellipsoid (eg Just 1981, Kvapil 1992, Rustan 2000). Its shape is a function of the distribution of particle sizes within the flowing mass and of the width of the discharge opening. The smaller the particle size, the more elongated is the ellipsoid of motion for the same discharge opening width. The upper portion of the ellipsoid of motion tends to be flattened or broadened with respect to a true ellipsoid and be shaped more like an inverted tear drop, particularly for large draw heights and irregular particle sizes. Furthermore, as will be discussed and illustrated in Chapter 7, more recent studies (Gustafsson 1998) have shown that for coarser, angular and gap-graded materials, more irregularly shaped flow patterns may develop. Nevertheless, for ease of calculation and explanation, the original flow ellipsoid theory will be developed here. The shape of a given ellipsoid of motion can be described by its eccentricity ε=

(

1 a N2 - b N2 aN

)

1/ 2

(6.1)

where aN and bN are the major and minor semi-axes of the ellipsoid of motion as shown in Figure 6.9. The composition of the ore and its particle size distribution, mechanical properties and moisture content will all influence the shape of the ellipsoid and its eccentricity. Smaller

257

Chapter 6: Extraction Level Design

particles will produce a larger eccentricity than larger particles. The eccentricity may also be influenced by the rate of draw with a higher rate of draw producing a smaller draw width and therefore higher eccentricity (Otuonye 2000). It is assumed that all horizontal cross-sections of the ellipsoid are circular although this has now also been questioned with an elliptical cross-section being suggested (Laubscher 2000). Janelid and Kvapil (1966) suggested that, in practice, ε varies between 0.90 and 0.98 with values in the range 0.92 to 0.96 being most common. If EN is the volume of material discharged from an ellipsoid of motion of known height, hN, then the corresponding value of the semi-minor axis of the ellipsoid can be calculated as

bN

⎛ EN = ⎜⎜ ⎝ 2.094hN

⎞ ⎟⎟ ⎠

1/ 2

(6.2)

or

bN =

(

hN 1- ε2 2

)

1/ 2

(6.3)

For this ellipsoid there will be a corresponding limit ellipsoid of volume EG, beyond which the material remains stationary. The material between the boundaries of the two ellipsoids will loosen and displace but will not report to the discharge point. Janelid and Kvapil (1966) represent this loosening factor as

β=

EG EG - E N

(6.4)

They found that β varies between 1.066 and 1.100, but that for most broken ores, β tends towards the lower end of the range so that EG ≈ 15 EN

(6.5)

Assuming that the limit ellipsoid has the same eccentricity as the ellipsoid of motion, equations 6.2, 6.3 and 6.5 can be used to calculate its height as hG ≈ 2.5 hN

(6.6)

258

Chapter 6: Extraction Level Design

As material is discharged progressively, the sizes of the ellipsoid of motion and of the corresponding limit ellipsoid, continue to grow. A useful dimension in the design of caving layouts is the radius of the limit ellipsoid at the height hN (Figure 6.9),

[

(

r = hN (hG - h N ) 1 − ε 2

)]

1/ 2

(6.7)

In practice, for the range of assumptions made and equations 6.1 to 6.6 given by Janelid and Kvapil (1966), r ≈ bG as shown in Figure 6.9. Otuonye (2000) gives a further useful equation that may be used to calculate the volume of the limit ellipsoid, EG, from a measurement of the ellipsoid’s width, u, at any height, h:

⎡ 1 ⎛⎜ u 2 + w 2 2 ⎢ E G = πbG h ⎢1 - ⎜ 2 3⎝ 4bG2 ⎢ ⎣

2 ⎞ ⎛ ⎟ - ⎜1 - u ⎟ ⎜ 4b 2 G ⎠ ⎝

1

⎞2 ⎟ ⎟ ⎠

2 ⎛ ⎜1 - w ⎜ 4b 2 G ⎝

1 ⎤ ⎞2 ⎥ ⎟ ⎟ ⎥ ⎠ ⎥ ⎦

(6.8)

where w is the width of the bottom opening and the value of the semi-minor axis of the limit ellipsoid, bG, is known or estimated. It must be pointed out that the approximate relationships and parameters involved in this approach were established for sublevel caving and are now quite old. They require validation for the coarser fragmenting ores now being mined by block and panel caving methods. Some authors postulate cylindrical rather than ellipsoidal draw zones (eg McCormick 1968) but the design principles remain much the same in the two cases (eg Hustrulid 2000). For the very high draw columns now being used or proposed for some operations, the differences between cylindrical and ellipsoidal draw zones are minimal.

6.4.2

Drawpoint Spacing

Three different definitions of drawpoint spacing may be used. The first of these spacings is that between the draw zones within a drawbell worked from two sides as in the typical layouts illustrated in Figures 6.3, 6.4, 6.6 and 6.7. The second is the centre-to-centre spacing of drawpoints in a direction parallel to the production drift axes across the minor apex as in the cases of the 15 m spacings shown in Figures 6.3, 6.4 and 6.6. Unless otherwise indicated, it is this spacing that will be referred to in this section. A third spacing that must also be considered in design is that across the major apex, referred to frequently by Laubscher (1994, 2000). In the herringbone layout shown in Figure 6.3, this spacing is approximately 20 m. In the offset herringbone layout shown in Figures 6.4, the drawpoint spacing “on the diagonal” across the major apex is more like 22 m. Establishing the “correct” spacing of the drawpoints requires a careful analysis of the interactions of several factors including the fragmentation and flow characteristics of the ore

259

Chapter 6: Extraction Level Design

(including the changes that occur with continuing draw), the method and planned rate of draw, geotechnical and design factors influencing the strengths of pillars, and production and cost considerations including the numbers of drawpoints required to achieve the required levels of production. Obviously, as the drawpoint spacing increases, the number of drawpoints that must be developed and the cost of development will decrease. Increasing drawpoint spacing will also increase the sizes of the pillars between drawpoints and allow larger drifts and equipment to be used in an attempt to increase the efficiency of production. However, the drawpoint spacing is also a function of fragmentation. The ellipsoid of draw concept can provide the basis for the selection of an initial spacing. The application of this approach requires a knowledge of the shapes and dimensions of the ellipsoid of draw and/or the limit ellipsoid. This knowledge may be obtained from model experiments (eg Heslop and Laubscher 1981) or from measurements made in full-scale trials or operations (eg Alvial 1992, Gustafsson 1998, Laubscher 2000). Figures 6.10 and 6.11 taken from an account given by Richardson (1981), illustrate the principles involved for the case of a draw zone with a circular cross-section. If the draw zone has a cross-section of another shape, possibly an ellipse, similar but slightly more complex considerations will apply (Laubscher 2000). In the case shown in Figure 6.10a, the draw zones for adjacent drawpoints do not overlap, isolated draw zones develop and pillars of undrawn ore are left between the drawpoints. This results in the potential loss of ore and in the application of excess weight to the major apex with the associated potential for damage to the extraction level excavations. On the other hand, in the case shown in Figure 6.10b there is a significant overlap between the draw zones. No ore is lost or additional loads imposed, but when the draw zone intersects the waste overlying the orebody, there is potential for waste to be pulled down between the drawpoints resulting in dilution. This effect can be exacerbated if the waste comminutes readily and can flow more easily than the ore (Richardson 1981).

(a)

(b)

Figure 6.10: Idealized vertical section showing (a) excessive drawpoint spacing with non-overlapping draw zones, and (b) close spacing with overlapping draw zones ( after Richardson 1981) 260

Chapter 6: Extraction Level Design

The conclusion to be reached from these simple considerations is that the drawpoint spacing should be such that the draw zones just overlap. Thus, as a first approximation, the drawpoint spacing should be slightly less than twice the value of the semi-minor axis of the limit ellipsoid, bG, shown in Figure 6.9. However, in block and panel caving mines, the drawpoint width, w, or more correctly, the active drawpoint width, wa, will have a finite value which should be taken into account. This is done by taking the drawpoint spacing required for adjacent draw zones to just overlap as S = wa + 2 bG

(6.9)

A further consideration, that of the layout of the drawpoints in plan, is illustrated in Figure 6.11. If the cross-sectional area of the draw zone is circular, the best “theoretical” arrangement is for the drawpoints to be placed on a hexagonal pattern as shown in Figure 6.11a as this minimises the amount of undrawn ore left between draw zones. Figure 6.11b shows a square pattern which produces comparatively greater zones of undrawn ore than a hexagonal arrangement with the same spacing. Reducing the drawpoint spacing to reduce these “dead” zones, has other undesirable consequences in terms of increased development costs, reduced pillar sizes and operational inefficiencies.

(a)

(b)

Figure 6.11: Idealised cross-section showing (a) hexagonal, and (b) square drawpoint spacings (after Richardson 1981) Figure 6.12 illustrates the application of this simple approach to the four main types of extraction level layout discussed in Section 6.3. For purposes of comparison, it is assumed that the draw zones have a constant diameter of about 16 m throughout so that adjacent draw zones just overlap across the minor apices for each type of layout. Clearly, the pair of draw zones generated by the two drawpoints served by the same drawbell overlap to a significant extent. In addition to the advantages and disadvantages discussed in Section 6.3, Figure 6.12 shows that the El Teniente and, to a lesser extent, the offset herringbone layouts have advantages over the other layouts in that they leave smaller “dead” zones between draw zones across the major

261

Chapter 6: Extraction Level Design

apices. However, in practice the situation is not as simple as that shown in Figure 6.12. Figure 6.13 shows the progressive development of draw zones in an offset herringbone layout as draw progresses from an isolated drawpoint, to two drawpoints serving a drawbell, to adjacent drawbells and then to the next line of drawbells.

Figure 6.12: Idealised cross-section showing draw zones for four major types of extraction level layout Flores (1993) describes the application of this approach as presented by Kvapil (1992) to the development of a revised mining plan for El Teniente 4 South using mechanised panel caving. In this case, experience with existing layouts could be used to obtain the parameters required to calculate drawpoint spacings. For example, Flores (1993) was able to estimate the value of the eccentricity as 0.96 for both the draw and limit ellipsoids. Since the height of the panel concerned was 260 m, the height or major axis of the ellipsoid of draw or extraction ellipsoid, hN, was estimated from earlier measurements to be one quarter of the panel height or 65 m so that the major semi-axis, aN, was 32.5 m. From equation 6.6 the height of the limit ellipsoid or ellipsoid of loosening, hG, is 2.5 hN or 163 m and its semi-major axis, aG, is 81.5 m. Given that ε = 0.96, and aN = 32.5 m, the value of bN can be calculated from equation 6.1 as 9.25 m. Thus, the diameter of the extraction, dE or the drawpoint spacing for touching draw zones, is given by equation 6.9 as 21.5 m with the effective width of the drawpoint (or the undercut in the specific case analysed by Flores), wa, taken as 3 m.

262

Chapter 6: Extraction Level Design

Figure 6.13: Progressive development of draw zone interaction in adjacent lines of drawbells (Laubscher 2000)

263

Chapter 6: Extraction Level Design

Figure 6.14 shows the draw or extraction and limit ellipsoids superimposed on the undercut layout being used before it was revised. There is no overlap of the extraction ellipsoids which means that passive pillars are left between the undercut drifts imposing high loads on the extraction level development below. In the design proposed by Flores (1993), the initial geometry was revised to that shown in Figure 6.15.

Figure 6.14: Gravity flow parameters and undercut geometry, El Teniente 4 South (Flores 1993)

264

Chapter 6: Extraction Level Design

Figure 6.15: Gravity flow parameters and proposed revised undercut geometry, El Teniente 4 South (Flores 1993)

In the preceding paragraphs, the classical ellipsoid of draw and limit ellipsoid concepts were presented and applied to the determination of centre-to-centre drawpoint spacings. However, as will be discussed in Chapter 7 and illustrated in Figures 7.1 and 7.2, the flow of broken material during draw can be more complex than assumed in this simple model. In particular, the shapes and sizes of the “ellipsoids” are greatly influenced by the shapes, sizes and grading of the fragmented particles. In many cases, there may be inter-mixing of material between the draw zones of adjacent drawpoints even when the theoretical ellipsoidal draw zones do not overlap in the manner illustrated in Figures 6.10 and 6.11. The diameter of the draw zone for an isolated drawpoint, or the isolated draw zone (IDZ) diameter, D, is a function of the size distribution of the broken rock. A common rule of thumb used in layout design is that interaction can generally occur if the drawpoint spacing is less than 1.5 D (Laubscher 2000). Figure 6.13 illustrates how interaction can be developed progressively in a group of drawpoints.

265

Chapter 6: Extraction Level Design

Drawpoint spacings of 15 m as illustrated in the examples shown in Figures 6.3. 6.4, 6.6 and 6.7 have been commonly used in modern block and panel caving mines. However, with some of the newer mines and the newer parts of established mines operating in more coarsely fragmenting ores, drawpoint spacings have increased. For example, the spacing at Northparkes E26 Lift 2 will be 18 m (Duffield 2000) and that at Palabora 17 m (Calder et al 2000). As was noted at the beginning of this Section, in practice drawpoint centre to centre spacings are not always on a square plan. There is usually a larger spacing across the major apex than across the minor apex. Figure 6.19 to be introduced below shows examples of spacings of from 14 m x 15 m to 15 m x 20 m used in the different sectors of the El Teniente mine.

6.4.3

Drawpoint Size, Shape and Orientation

From a stability viewpoint, the drawpoint and the drawpoint drift should be as small as possible. In practice, drawpoint size will be determined by the particle sizes of the broken ore and the sizes and operational requirements of the loading equipment which, in turn, should be determined by the characteristics of the ore. The blocking of drawpoints by oversize blocks should be avoided but cannot always be guaranteed. The percentage of broken ore reporting as blocks of more than 2 m3 in volume is often used as an indication of the propensity of the orebody to produce large fragments (Laubscher 2000). (This size is equivalent to a cubic block having a side of 1.26 m.) It has been suggested on the basis of model experiments and field experience that the drawpoint size should be three to six times that of the largest fragment (Kvapil 1965, Oyuonye 2000). In practice, modern drawpoint drifts in mechanised layouts in the stronger orebodies are typically 4 m or more in both width and height with the effective sizes of the drawpoints themselves being smaller. Drawpoint drifts and drawpoints are usually of the typical shapes of drifts in metalliferous mining with rounded top corners and possibly roofs. In most circumstances, these curved shapes are more stable than flat roofs especially as spans increase. They are also easier to construct using drill and blasting techniques. Furthermore, the shotcrete lining and steel set support sometimes used at the drawpoint are more conducive to the curved shape. However, as illustrated in Figure 6.16, drawpoints have been constructed with flat roof sections, most notably in what is sometimes known as the Henderson design. The orientation of the drawpoint brow with respect to the major and minor discontinuities present in the rock mass is an important factor in determining the stability of the brow. Figure 6.17 illustrates the point for an idealised case in an offset herringbone layout in a rock mass containing one or more sets of relatively steeply dipping discontinuities striking in a preferred direction. For the relative orientations of discontinuity strike and the excavations shown on the left hand side of Figure 6.17 the discontinuities strike perpendicular to the plane of the drawpoint brow. The horizontal or sub-horizontal stresses induced across the drawpoint brow will tend to clamp the discontinuities in place and minimise the amount of slip and block fallout that is possible. In conventional rock engineering analyses (eg Hoek and Brown 1980), this

266

Chapter 6: Extraction Level Design

orientation would be regarded as unfavourable in terms of production drift stability although this is not likely to be the critical design issue in this case.

Figure 6.16: Concreted drawpoint with flat roof (Butcher 2000b)

On the other hand, the orientations shown on the right hand side of Figure 6.17 are more favourable for production drift stability but are relatively unfavourable for drawpoint brow stability. The discontinuities strike parallel to the plane of the brow and parallel to the stresses induced across the brow. This means that they will not be clamped by the stresses and that much larger blocks will be free to slide or fall from the brow than in the former case.

267

Chapter 6: Extraction Level Design

Figure 6.17: Illustration of the influence of discontinuity and excavation orientation on drawpoint brow stability (after Laubscher 2000) 6.4.4

Drawbell Geometry

Laubscher (2000) points out that the term drawbell is descriptive in that the ideal shape of of a drawbell is presumably like an inverted bell in order to obtain the best possible flow of ore to the drawpoint. Improved ore flow is often reported with shaped drawbells. However, there must be a compromise between the shape of the drawbell and the strengths of the major and minor apices which must remain stable throughout the life of the drawbell and drawpoint. As well as the two primary factors of ore flow and pillar strength, the drawbell shape will also be influenced by the undercut design and the practical drilling and blasting issues discussed in Section 5.6.5. Figure 6.18 shows sections through the drawbells across the major apex for three possible designs. The three cases illustrated show how inclining the drawbell sides should produce better flow characteristics than vertical sides and a large, flat top to the major apex. In the design illustrated in Figure 6.18a, poor interaction of the draw zones is obtained and production is likely to be lost. The issue of ore stacking discussed in Section 5.4.3 will also arise. In the second case illustrated in Figure 6.18b, the sides of the drawbell are inclined with the other design parameters remaining constant. The interaction obtained between draw zones should be improved over that in the first case and the extraction of ore should be more efficient. There is, however, a decrease in the size of the pillar. A further improvement to the flow characteristics may be achieved by using an inclined undercut with an increased slope and a corresponding increase the overall drawbell height as illustrated in Figure 6.18c. The strength of the pillar is further reduced in this case and would have to be evaluated before such a design was adopted.

268

Chapter 6: Extraction Level Design

Figure 6.18: Influence of drawbell shape on ore flow and pillar shape (after Laubscher 2000) As illustrated in Figures 5.28 and 5.29, the shapes of drawbells in plan or horizontal crosssection can be quite complex in modern designs. In the El Teniente layout as illustrated in Figure 6.7, for example, a number of choices exist for the orientation of the drawbell sides with

269

Chapter 6: Extraction Level Design

respect to the production and drawpoint drifts. Figure 6.19 shows a number of designs that have been used in a range of particular circumstances at El Teniente, mainly in the Teniente 4 South sector (Jofre et al 2000).

Figure 6.19: Some drawbell shapes used at El Teniente (Jofre et al 2000)

6.5 6.5.1

SUPPORT AND REINFORCEMENT Terminology

Because of their vitally important roles in maintaining production and the high and changing stresses imposed on them, much attention has been traditionally paid to the support and reinforcement of extraction level excavations. Before discussing the issues involved and the

270

Chapter 6: Extraction Level Design

techniques used, it is important to clarify some aspects of the often confused and loose terminology used in this area. The terms support and reinforcement have been used throughout this book and are defined in the glossary given in Appendix A. The definitions adopted distinguish between support and reinforcement and are due to Windsor and Thompson (1993) and Windsor (1997). Support is the application of a reactive force at the face of the excavation and includes techniques and devices such as timber, fill, steel or concrete sets or liners and shotcrete. Reinforcement on the other hand, is a means of improving the overall rock mass properties from within the rock mass by techniques such as rock bolts, cable bolts and ground anchors. Reinforcing elements may be tensioned or untensioned on installation. In the latter case, they are referred to as dowels. Support or reinforcement may also be described as being either active or passive. Active support or reinforcement imposes a predetermined load to the rock surface at the time of installation. It may take the form of tensioned rock bolts or cables or, in other applications, hydraulic props, powered supports or segmented concrete linings. Passive support or reinforcement is not installed with an applied loading but develops its loads as the rock mass deforms. Passive support or reinforcement may be provided by steel arches, mass concrete or untensioned rock bolts, reinforcing bars or cables. The timing of the application of support and reinforcement can have a major influence on its effectiveness. Of particular interest in the current context is the concept of pre-reinforcement which is the application of reinforcement prior to the creation of the excavation. This applies some constraint to deformation and increases the rock mass strength before the loosening often associated with excavation can occur. On the other hand, post-reinforcement is the application of reinforcement at an appropriate time after the creation of the excavation. As the discussion of undercutting strategies in Section 5.2 should indicate, the post-reinforcement or post-support of extraction level excavations will be too late in many circumstances. An alternative approach to terminology developed for the rock burst conditions encountered in the deep level gold mines of South Africa ( Ortlepp et al 1999, Stacey and Ortlepp 1999, 2000) is useful for some applications in caving mines, particularly where large amounts of deformation and damage, or even rock bursts (see Chapter 10), are induced. Here, the support and reinforcement systems used may be described as being either retainment or containment support. In this terminology, retainment support refers to those elements and systems which act within the rock mass to reinforce it for some distance in from the excavation boundary in the same way as reinforcement in Windsor and Thompson’s terminology. Containment support refers to those elements and systems such as wire mesh, straps, lacing, shotcrete and other sprayed membranes, used to contain the broken rock mass around the excavation.

271

Chapter 6: Extraction Level Design

6.5.2

Principles

The principles of good support and reinforcement practice have been given in a range of textbooks (eg Brady and Brown 1993, Hoek and Brown 1980, Hoek et al 1995) and other publications (eg Hoek 2001, Windsor 1997, Windsor and Thompson 1993). The principles to be outlined here apply to the provision of support and reinforcement for extraction level excavations which, as has been established in Chapter 5, are subject to high induced stresses and susceptible to rock mass failure. The continuity of production operations on extraction levels is so critical in block and panel caving mines that a conservative approach to the provision of support and reinforcement is often adopted. While this approach can be effective in ensuring that the excavations remain stable throughout their design lives and, in some cases require no repair, it does not represent best practice because it may involve more expenditure on extraction level installations than is absolutely necessary. The general principles of good support and reinforcement practice may be illustrated through ground-support interaction, characteristic line or convergence-confinement concepts as in the well-known diagram shown in Figure 6.20. Although expressed in terms of “support”, these concepts also apply to what is referred to here as reinforcement. The details of this approach have been given elsewhere (eg Brady and Brown 1993, Hoek and Brown 1980, Hoek et al 1995, Hoek 2001) and will not be repeated here. The essential feature of this approach as illustrated in Figure 6.20, is that it shows how the support or reinforcement helps mobilise and conserve the inherent strength of the rock mass surrounding the excavation even when it is in a yielded or broken state. The support or reinforcement helps the rock support itself. This approach also illustrates clearly the importance of the timing of installation and the stiffness and yield characteristics of the support and reinforcing elements. A practical example of the use of this approach in establishing extraction level support requirements given by Lorig (2000) and Leach et al (2000) will be summarised in Section 6.5.6.

Figure 6.20: Simplified ground-support interaction diagram for a circular tunnel excavated in a hydrostatic stress field (Hoek 2003) 272

Chapter 6: Extraction Level Design

Based on considerations of ground-support interaction mechanics, Brady and Brown (1993) developed a set of general principles for good support and reinforcement practice. These principles were not meant to apply to the case of providing support for the self-weight of an individual block of rock, but to the more general case in which yield of the rock mass surrounding the excavation may be expected to occur. They do not refer specifically to the provision of containment support which will always be required in the application being considered here. These principles as adapted by Brown (1999) are reproduced here with further modification and expansion as necessary to take into account the special features of extraction level installations in block and panel caving mines. It is interesting to note that despite their differing origins, these principles have many points in common with the guidance offered by Laubscher (2000). In this list of principles, the word support is used throughout in the interests of simplicity but should be taken to refer to both support and reinforcement as defined here. 1.

Install the support close to the face soon after excavation. In some cases, it is possible and advisable to install the support (or, as is more likely, the reinforcement) before excavation or before the excavation is complete. In others, usually involving high “squeezing” pressures, it may be advisable to permit some displacement to occur before the support is installed.

2.

There should be good contact between the rock mass and the support. If this is not achieved, it has the effect of reducing the effective stiffness of the support in which case excessive displacement and loss of rock mass strength may occur.

3.

The deformability of the support should be such that it can conform to and accommodate the displacements of the rock mass. This means that in high stress and dynamic loading environments, the support should be capable of yielding and retaining load carrying capacity.

4.

Ideally, the support system should help prevent deterioration of the mechanical properties of the rock mass with time due to weathering, repeated loading or wear. The application of this principle to the support and reinforcement of drawpoints will be illustrated and discussed in Section 6.5.5 below.

5.

Repeated removal and replacement of support elements should be avoided.

6.

The support and reinforcement system should be easily adaptable to changing rock mass conditions and excavation cross-section. This principle does not assume the importance in block and panel caving extraction level design that it does in other mining methods because the rock mass conditions will generally be reasonably constant so that the same design can be used for a given block or panel.

273

Chapter 6: Extraction Level Design

7.

The support and reinforcing system should provide minimum obstruction to the excavations and the working face. In the case of caving mines this principle can be extended to account for the need for the support and reinforcement system, and particularly the containment support system, to be able to withstand impact and abrasion by loading equipment and blocks of rock.

8.

The rock mass surrounding the excavation should be disturbed as little as possible during the drill and blast excavation process so as to conserve its inherent strength.

A further ground support principle that was not included in Brady and Brown’s list but is well known in civil engineering tunnelling practice (eg Brown 1981, Kovari 2001), is that under high stress conditions, support and reinforcement performance can be improved by “closing the ring” of shotcrete or a concrete lining across the floor of an excavation. In modern mechanised block and panel caving mines, providing and maintaining a good concrete floor on the extraction level is important in controlling costs and maintaining the efficiency of the extraction operations. This brings with it the opportunity to “close the ring” by making the concrete floor integral with the overall support and reinforcement system as illustrated in Figure 6.16.

6.5.3

Support and Reinforcement Elements

The support and reinforcement elements and systems used on the extraction levels of block and panel caving mines will be familiar to most readers of this book. They will be listed and briefly described here for completeness. Fuller details of most of them are given by Hoek et al (1995), Laubscher (2000), Ortlepp (1997) and Wilson (2000).

Rock bolts consist of an anchorage, a shank, a face plate, a tightening nut and, in some cases, a deformable bearing plate. By definition rock bolts are tensioned and they may be cement or resin grouted along their lengths. In the applications being considered here, the practical lengths of rock bolts are usually restricted by space limitations in the excavations from which they are installed. Dowels, usually consisting of grouted reinforcing bar, have been widely used for reinforcing extraction level excavations as in the example shown in Figure 6.26. Cable bolts can be longer than rock bolts and can sustain much greater axial loads. For the best results they should be grouted and tensioned, although untensioned grouted cables have also been used. Cable bolts are especially useful for reinforcing the large spans created at excavation intersections as illustrated in Figure 6.31 below. Wire mesh usually consisting of 4 mm diameter wire either welded on a 75 or 100 mm square pattern or in the form of chain link mesh, is used to contain broken surface fragments and as reinforcing with shotcrete.

274

Chapter 6: Extraction Level Design

Shotcrete or sprayed concrete is used to provide passive support to the rock surface and forms an integral part of most support and reinforcement systems used in modern mines. It may be applied in several layers of tens of centimetres in thickness using either a wet-mix or a dry-mix process. The former process is the most common in modern underground mining. Shotcrete may be reinforced with wire mesh of with steel or other fibres to improve its toughness, durability, shock resistance, and shear and flexural strengths. Sprayed membranes of other types are being used increasingly in metalliferous mining instead of mesh and for other special purposes (Spearing and Champa 2000). Because of the high stresses and the need for absolute security in extraction level excavations, it is considered unlikely that they will find much use in block and panel caving, at least until their mechanical properties have been improved considerably. Mass concrete has long been used as a support element in block caving mines (eg Butcher 2000c, Gallagher and Loftus 1960). In the more traditional block caves in weak rocks, mass concrete could be expected to be stronger than the rock mass that it replaced. This cannot always be expected to be the case in modern mines in stronger rock masses. In these cases, mobilising and conserving the inherent strength of the rock mass using other support and reinforcement systems and the principles outlined in Section 6.5.2 is to be preferred. Straps and lacing made from flat steel plate, tendons or mine rope have been widely used to contain the surface rock between rock and cable bolts and the rock at the acute (bull nose) and obtuse (camel back) angles at the intersections of production and drawpoint drifts. Wilson (2000) gives descriptions of a range of these measures used at the King and Shabanie mines, Zimbabwe, including a technique known as rock stapling which uses steel cable or rope. Steel arches were used extensively in the early days of block caving and, as illustrated in Figure 6.16, are still used especially in drawpoints and in conjunction with shotcrete. Experience in block caving and other forms of mining has shown that, when used alone, steel sets are not as effective as might be supposed unless the principles of good support practice outlined in Section 6.5.2 are followed. The Toussaint-Heintzmann yielding arch which includes elements which yield at constant load have been found useful in controlling large deformations in some cases. 6.5.4

Stress – Strength Analyses

As in other forms of underground rock engineering, stress-strength analyses may be used in design to study the likely responses of extraction level excavations. Examples of the two dimensional calculation of the elastic stresses induced on the boundaries of extraction level drifts under a range of scenarios were given in Chapter 5. However, because of the truly three dimensional nature of the problem, the stress paths followed by the rock on the boundaries of excavations through their lives, and the likelihood that plastic deformation of the rock mass will

275

Chapter 6: Extraction Level Design

occur, the situation in practice is likely to be more complex than that shown in these analyses and more difficult to replicate numerically. Because of the difficulties involved in carrying out the necessary stress-strength analyses in many cases, recourse is often made to precedent practice and practical guidelines in extraction level support and reinforcement design. For example, Trueman et al (2002) describe the use of a method based on the Q system of rock mass classification introduced in Section 2.7.3. A simpler and more preliminary assessment of likely support requirements in a drift may be made using Table 6.2 which shows approximate tunnel support design categories for a range of rock mass strength to in situ stress ratios. The use of even this simple approach requires a knowledge of the in situ stress regime and an estimate of the rock mass strength that is usually made using the Hoek-Brown criterion (Hoek and Brown 1997).

Table 6.2: Approximate support design categories (Hoek 2003) Rock mass strength/

Support category

in situ stress > 0.5

No serious problems anticipated; support usually chosen to deal with local safety issues; detailed support design studies not necessary Routine support design usually adequate; rock bolts or shotcrete or

0.3 to 0.5

light steel sets are normally adequate. Relatively simple design approaches, such as rock support interaction using ground reaction curves, are normally adequate

0.17 to 0.3

Careful design of support required; reinforcement (rock bolts) with mesh- or fibre-reinforced shotcrete required Serious instability anticipated; very detailed design of support

0.05 to 0.17

required; yielding steel sets may be used; numerical analysis is advisable

30 MPa) ahead of or directly below the undercut face. Deformation of the drift walls increases rapidly when the support pressure falls below 6 MPa. This level of support pressure is impractical to supply and any drift development in this zone would require substantial yielding support and reinforcement.

•

Moderate stress conditions (20-30 MPa) where production drifts are effectively less than 10 m in plan from the edge of the undercut. Deformation increases rapidly when the support pressure falls below 2-3 MPa. Again, this level of support pressure is probably not feasible with conventional techniques but may be achievable with mass concrete placed once the undercut is advanced.

•

Lower stress conditions (